Sometimes you read something that stays with you long after you’ve returned the book to the library. Not just the author, or the title, but a quote from the book or a concept that flits around in the back of your head when you least expect it.

I started my professional life, if one can call working at age 16 a professional life, as a page in a city library. A page is the job title for someone who returns books to their rightful place on the shelf. It’s a brilliant play on words, and also easier to say then “book reshelver.” It’s also where I got to turn my love of reading into a paid job. It also taught me a lot of skills that I’ve relied on later in life. The first and most closely tied to AI is the concept of a taxonomy. But that’s a subject for another time. Today is about books.

There are always lists of books – books that should be banned, books you should read, books that are recommended by schools, books that are recommended by book award groups. There are books about burglars (see Fiction, or section 364: Criminology & True Crime), books about dogs (see 599 for canines or 636.7 for books on dog breeds), and there are books about ships (again we stay in Applied Sciences with 623.8).

So what then do burglars, dogs and ships have in common. It feels like the start to a bad joke, so I asked Gemini. And suggested it “think outside the box.” I got back the concepts of “barks, board and watch.” All with explanations on why it works. I then said “What if the common thread was Spinoza?” Gemini gave me two options. I favored the one that used the phrase “A brilliant curveball connection!”

Spinoza – see 199 for Dutch Philosophy / Modern Western Philosophy in most public libraries; though some put it in 170 - Ethics

When I was trying to think of what to call this article the title “On Human Bondage” came into my head – from the Modern Library’s list of best English-language novels. It was number 66 on the list – but this is about 1927 and the book by Maugham was published in 1915 [823.912]. The novel takes its title from Baruch Spinoza’s “Ethics: On Human Bondage, or the Power of the Affects.” Which gets us closer to where I was going with the thought process on the title. Which then lead me to recall a book I had recently shelved on my own bookshelves by Lawrence Block “The Burglar Who Studied Spinoza.” If you like crime novels and mysteries, check out Lawerence Block’s Bernie Rhodenbarr. I’d recommend him to readers of the “Cat Who” mysteries (Lilian Jackson Braun) or any lover of “cozy mysteries” (the ones without all the gore and details).

But, back to the point. My tangent onto Spinoza wasn’t wasted as I was looking him up, because my Modern Western Philosophy class was years ago and Spinoza’s not mentioned in Bruce’s Philosophers Song from Monty Python, so I had no musical clue on his general philosophy.

Essentially Spinoza defined that human morality is defined by good and evil, and our knowledge thereof. He referred to bondage as human’s inability to moderate their own emotions as we struggle though the concepts of passion being controlled by fortune (fate) rather than ourselves. We free ourselves when we can understand our place in nature and recognize the causes behind what we feel and why. Since you can’t remove emotion entirely from your life you are increased to act (joy) while sadness (evil) decreases our will…. So we struggle through this as external forces (others, the world – what Spinoza called Affects) impacts us. Essentially we are tossed around like ships in a storm…

Basically – when a man is prey to his own emotions – he is not his own master. And that brings us to Dogs and my initial commentary about a book sticking with you.

It took me a long time to find the title and author of the book, and to be honest, it took a lot of AI searching, despite the Goodreads group that helps individuals find books based on vague descriptions. I won’t bore you with the prompts.

The book is “The New Creatures” by Mordicai Gerstein. It was published in 1984, so it was already almost 10 years old when I discovered it on the library shelves as I was reshelving books in the children’s section. If you don’t want spoilers – stop reading now as I start to relate how this book and AI are singularly linked in my mind.

As AI began to spread across my current job role, I began to try and recall this book even harder. The similarities in the plot line were unnerving to me. And as my colleague Ida Helena Rust asked the question of “When AI Becomes a Good, Will You Have Faith?” it really hit home.

The concept of the book begins with a child commenting that “dogs and cats are nice but it seems all we do is wait on them!” And so his grandfather is off on a tale about when the world was ruled by cats and dogs. But cats and dogs were lazy – as much as they are now – but couldn’t find anyone to wait on them. So… a Dog (thus cementing the concept that dogs are smarter than cats) set off to find a “creature” to wait on them. He finds this “new creature” – which is very smart and trainable….

Just like AI is trainable… and getting smarter every day….

And thus – that is how humans came to rule the world over cats and dogs. And now humans are becoming more and more lazy, and AI is becoming smarter and smarter… and concepts such as physical AI are starting to take shape. Robots are being endowed with AI, and that AI could become able to “think” for itself…. So, I worry – will humans become the next cats and dogs? [See 629 Other branches of engineering for Robots]

Unfortunately, I see it happening. As my husband job hunts for a new software development job, he is often told in interviews that there will be a coding challenge that will either be graded by AI, which will then be reviewed by the developers, or which portions of the test he can use AI and which he can’t – or my favorite yet: They want to install a tracking software on his computer – before he’s even hired” to see if he uses AI to complete the test. [The use of AI not being wrong, but apparently they want to track how much is put into AI to determine how much you can actually do on your own.] All I can think of is when I used to sit and type up the Dean of Engineering’s papers for journals because he couldn’t type and I had to transcribe things. Nothing says you couldn’t put the problem into AI on your phone or another computer…. So is this actually accomplishing anything?

Does it matter when kids are being taught to prompt AI rather than look things up in a card catalog? [See Computer Science, Information and General Works 000, 006.3 specifically for AI (006 being specialized computer methods) Although to be more precise in academic libraries – see Q334-336 with AI being under Q: Science and Cybernetics, though if you want to find algorithms – see QA76.9 or Q76.87 for neural networks – related to computers. Related to the brain refer to QP351-495 or BF311 (Cognitive Neuroscience/Psychology)]

Does it matter if public schools, which are supposed to teach facts and science are being pushed to teach a specific narrative by selected humans? Is it any different than if AI were to narrate the conversation? Perhaps – AI is supposed to be agnostic in it’s beliefs, though biases exist. If the government was run by AI would it be any better or worse than it is now – worldwide? I would hope that AI, having been trained on history would at least learn from the mistakes of the past and not predicate them forward.

Amusingly, about the time I discovered Gerstein’s book, The Simpsons aired an episode that has a line that is often parodied. In February 1994 (exactly 10 years after the book was published and five seasons into The Simpsons) Kent Brockman says “I, for one, welcome our new insect overlords.” [Insects 595.7] That phrase has since been misquoted as “Robot overlords,” “Alien overlords” and “AI overlords.” [Aliens 001.9 Fringe Science or 999 Extraterrestrial worlds]

Spend any time on social media or the internet and inevitably you run into an article about “The Simpsons Were Right About…” which then proceeds to explain how the cartoon series predicted something that happened. [See 300: Social Sciences, Sociology and Anthropology]

Will we eventually have AI overlords? I certainly hope not. [See 150 Psychology]

Which brings me back to the question of what burglars, dogs and ships have in common. I probably could have spent more time prompting AI to come up with better concepts than what it gave me back in the initial response. And even then the response it did give me had a hard time working in Spinoza beyond the title of Block’s novel. And I’ll admit that relating the emotional roller coaster humans endure to being a ship in a storm is a stretch, even for my brain – which was looking for some sort of alliteration for the title but couldn’t bring myself to call this “On AI: Bitches, Boats and Burglars” (they had to be in alphabetical order). [See 400: Language]

Now why have I included shelf numbers in this article? Because it’s part of a universal taxonomy that allows anyone to find something in a library worldwide – regardless of the language. Religion will always be in the 200s, 900s will always be history and the 700s will always be Arts and Leisure. Now, while it can be argued that the Dewey Decimal system is not being used as much as it once was and the one developed by the Library of Congress is taking precedence given its ability to break the concepts down further – to me it doesn’t really matter. The concept is the same. An ordered way to make information findable universally. With strict rules that mean things are usually in the same place anywhere in the world.

But Elizabeth, you said Spinoza could be under 170 or 199. You’re right. But let’s face it the difference between those two is not as great as if it was filed under Literature in the 800s. So, while thee is some determination being made by humans as to where things should be filed, the rules of a taxonomy are fairly clear – and this is one thing that makes me less nervous about AI. Somewhere, deep inside is a taxonomy that actually governs the “brain.” And while it may not always be evident, I don’t fear that taxonomy disappearing completely – at least not yet. In fact, a recent article in IEEE Spectrum delved into the original 1960s chatbot – ELIZA – which was really a series of personas with multiple scripts. Find the book the article references at 006.35 for Natural Language Processing/Communication.) I’m not sure that makes me feel any better or worse about where AI is going.

Do I think that AI still needs guardrails and that humans should choose not to rely on it as much as we are starting to – yes I do. Because I see
The New Creatures” book as a cautionary tale, much like any fable from Aesop. Does that stop me from hoping that I come back in my next life the fur baby of someone like me who is spoiled rotten and waited on hand and foot – no it doesn’t.

Because – returning to Spinoza’s ship… nothing calms a sea of emotion than the cuddle of a fur baby. The joy (read love) in their eyes helps calm us humans – and no matter how good AI or robots become – they will never be able to replicate the look from the ones we care about (human or fur baby). [Love – 152.41]

Find more about the brain behind this article in the 600s, specifically 612, Human Physiology.

At first, you recognize familiar constellations: The Great Bear, Orion, Scorpius… There are apps you can point toward the sky that tell you what you are looking at. They display the name of the constellation accompanied by an illustration of the mythological figure. Staring into the darkness through your smartphone. It even tells you about the stars that aren’t visible, yet.

You put your phone away and look up at the sky once again with your own eyes. Your neck begins to ache. You decide to lie down. At that moment, eager to see more, it no longer matters that the ground is cold or damp with evening dew.

Lying in the grass, dogs bark in the distance, suspicious rustlings here and there, daytime songbirds still chatter. As wide awake as guys in the pub.

Or perhaps you are lying on the sand. Waves roll on to the beach and retreat, making a whooshy sound. Shhhhh…. Shhhhhhjjjj…. Over and over and over again.

Stretched out on a city bench in a quiet corner of a big city. Amidst the remnants of artificial light, the stars stubbornly shine through. Sounds of passing cars fade into the distance. You couldn’t care less if someone happens to walk by. Absorbed in your own world. No fear, but the pain of this moment’s edges. The past and the future.

I am wearing a long winter coat with a fur-lined hood. It is big enough to keep my head cozy while in corps pose. I look at the sky. After a while, more and more tiny white specks become visible. Until the dark is not black anymore.

My breathing synchronizes with the sea. My sense of hearing is absorbed by the rushing sound. The discomfort of the cold weighs less than the peace this moment bestows.

Time passes. Eventually, I can no longer feel my hands and feet. The tip of my nose is frozen. My back hurts. The energy to raise myself up has vanished. I roll onto my side and like the puppeteer of a discarded wooden marionet, I reconstruct myself into a sitting position.

I already miss my outlook into eternity. Surrounded by darkness. An involuntary meditation on emptiness, covered by a fog of hopelessness, existing within a vacuum. Hopelessness stripped from hope that it will ever come to end.

This is the worst form: The absence of feeling. Fear, anger, sadness are emotions that already sit at the very lowest threshold of how you might wish to feel. At least, they are signs that you are alive, even in a state of utter misery. But when you no longer even experience those emotions, when the moment has devolved into a surrender, a becoming-one one with the idea of being “Dust in the wind” as the silent sound of The Scream, then you are frozen within a haunted painting.

The night is no friend to those who worry. Neither is the day. From behind thick glass, you observe the awakened world. A world to which you feel you do not belong, while at the same time suffering the intense pain of that separation. Day in, day out.

This tormenting feeling cannot be soothed by seeking out company or finding a satisfying activity. No one hears The Scream.

With your mouth shut, you continue your way. Every step leaves a trace of shame behind. You glance over your shoulder, but nobody has noticed. People smile at you, strike up an amicable conversation, and your mouth responds in the same tone.

On your way to work, to the gym, or during an elegant dinner at a restaurant. It’s all like water droplets sliding off a frictionless surface. They don’t touch you.

The only thoughts that stick are those about how to put an end to the suffering, the guilt of having this thought, and still having a faint glimmer of hope.

Where does this sense of guilt come from? To whom do you owe a debt? And what, precisely, do you owe? To the people who love you? But who is it that they actually love? Do they truly know you? Do you owe them your company? There might be the guilt you feel toward your employer and colleagues. Let’s be honest: in a corporate or organizational work environment, everyone is replaceable.

What if you possess a unique creative talent? It is often said that such talent does not belong to you; you are merely the medium through which this gift reaches the world. Many brilliant and successful people have chosen to leave this world on their own terms. Perhaps the stars were calling to them, or the foggy vacuum was suffocating them.

The night is the darkest when feeling has vanished.

The etymological dictionary states that ‘hope’ means ‘having confidence in the future’. ‘Confidence’ at its turn comes from ‘with’ ‘trust’. The glimpse of hope, smaller than the at first sight unseeable star in the dark night, that is always there. It is a trust accompanied by 99,99% distrust. If only you keep gazing long enough into the dark sky. And even then, you might still find yourself doubting your own senses.

It is this faint star in the unreachable distance that can guide you into the light. If only you find the strength to keep dragging yourself through the Valley of Death, thirsty and hungry. The light is extraterrestrial. “Carry on”. Incarcerated in Bir Barhout, human hands reach out toward you. To pull you out. To keep you from falling. If only you reach out towards these hands.

This hope, directed toward a star in the nearly invisible distance, is intrinsically human. A fraction of feeling left. Our capacity to feel. Imagine yourself weightless in an enclosed, dark space, devoid of any external stimuli. Your nervous system turns inwards. You perceive your heartbeat, the flow of blood through your veins, and other bodily sensations. That is you. Now, imagine suddenly feeling a wall, or hearing a sound, or seeing a light. All at once, there is a point of reference. You run your hand along the wall to discover where it leads. The same applies to the sound and light “Where is it coming from?”

Alternating between dragging yourself forward along a rocky, sandy path hemmed in by immense stone walls. Using your very last bit of strength. At least, you are feeling solid ground. Reaching your hand out from a deep, dark pit to be grasped by another. Experiencing human touch. It is ultimately this sensational connection, etymologically the ‘joining’ ‘together’, that saves you.

As long as you keep following the path, trusting in the faint light far off in the distance and grab the hands that reach out for you. Then you can fall. We won’t let you.

If you suffer from suicidal thoughts, please reach out.

Photos: Pexels

Prof. Venkatraman is Professor Emeritus at Boston University Questrom School of Business and author of The Digital Matrix and co-author of Fusion Strategy. In the session he challenged us to rethink how businesses, organizations, and individuals will create value in a world where human and machine intelligence increasingly work, learn, and evolve together.

• First-principles agentic intelligence: In agentic intelligence, human intelligence and machine intelligence start to come together framing problems jointly and then letting the machine solve the problem. Humans and machines co-create intelligence.

• The addition, multiplication and exponential framework:

  • Addition helps us work faster but not fundamentally differently.

  • Multiplication is when humans and machines start to interact.

  • Exponential is where humans and machines learn and co-create together.

• Economies of expertise: In the industrial age, economies of scale needed physical assets. In the intelligent age we have AI factories built to produce intelligence, which makes the old economies of scale less relevant. And the concept of agentic intelligence is about putting the economies of expertise into practice.

The discussion also dived into what this all means for society, education and all our futures. In this new age, how do we preserve human judgment, agency, and responsibility?

See what you think and let us know below!

It is dense! A whole video I ask you to watch. Several other links.

My recommendation - don’t stress!

The 4D Community is going to the Azores about a year from now. And the journey to the Azores goes through what we call a Slow Schoolhouse which is starting. So take your time.

Some thoughts:

• The video has some images and it is nice to watch as a video, but if you don’t have the time you can always listen to it like a podcast while commuting, driving, etc.

• The Erik Gandini documentary is fantastic. Watch it during a weekend when you have time. If you have teenagers, watch it together with them. A wonderful possibility to discuss what work means for them and what their future might look like.

• Finally, all the other links are given as possible references and in my attempt to connect the dots. Read them when you get a moment and if you have the interest.

Maybe the most important thing. The learning journey to the Azores will of course include AI. And among its many aspects, I think our own personal use of AI is obviously so important. Maybe that is the main message and what I wanted to reflect on, and why I invite you at the start to watch the video.

In this link - Azores 2027 - you can find more information about our learning journey towards the Azores, and also express your interest in participating.

Enjoy!

And now I have made this post a bit longer with this note, but it is with good intentions! 😊

I did this video on my very personal experience of AI, and how it is helping me a great deal since I retired and embarked in new projects.

It is already posted in the Playing the Second Half Youtube channel, and I wanted to post it here with some few additional thoughts.

So many articles and books and seminars on AI but I kind of see very little on personal reflections and use. So, as I repeatedly say in this video, that is what I am aiming at with that video.

But I would like also to bring some few additional thoughts – not captured in the video here in this post.

Please do however start with the video, if you have the opportunity, and come back to these three thoughts that I don’t either see so often covered in the AI debate.

1. The Nature of Work

In all debate about AI and particularly on AI taking away jobs, I never see any serious discussion on work itself, and why we need it or rather identify ourselves so much with it.

Maybe one exception is this fantastic documentary After Work, by Erik Gandini. In this link you find a short trailer and I would recommend watching the full documentary. It is some years old now but I find it timeless in many ways.

One particular moment that made an impression on me is when it highlights Gallup’s State of the Global Workplace report, prominently featuring the striking statistic that 85% of workers worldwide are not engaged or are actively disengaged at work. Imagine! And if you break that down, 18% actually hate their work!

So every morning, billions of people go to a job they either hate or are disengaged.

I just then have one question – why in all debate on AI taking away jobs, we don’t also discuss the nature of work and how possibly AI could change the dynamics into jobs that are more satisfying?

And lets be clear with one thing though. Even when we have awful jobs, when those disappear so it is a tragedy for the people involved and their families. Still I think, at least long term, that we should try to design and think of how an ideal society could look like instead of preserving the status quo. Because that also includes changing the conditions for the cobalt miners in Congo or the night shift workers in the US meat packing plants.

Would that not be wonderful? A Meta-Opportunity! For reference please see our podcast with Gaspar Wosa here, and how we should not let fear decide our future: What is the Metacrisis for you? And innovator’s perspective.

2. Authorship & Intellectual honesty

In the video, you will notice one thing I mentioned about being non-negotiable: authorship.

Something that is also reflected in previous posts by Elizabeth Roberts: An Identity Crisis and An Image Question.

What I find an issue is what I would call “Intellectual Honesty”. That happens when you see people very critical to AI that at the same time use heavily AI in what they write. To the point that the Substack AI detection tool shows 100% AI produced.

I don’t want to create social panic here but I find it so sad. Why not at least – as Elizabeth Roberts so well put – at least do a note or disclaimer? How can we then trust?

Finally, how do we ensure we are not outsourcing our thinking? How do we ensure, as we become more productive and empowered, that our minds are not being hacked?

Here, Alon Rozen’s brilliant post, Your Mind Is Not a Prompt, is a must-read and a powerful antidote to prevent exactly that from happening.

3. Unintended consequences of becoming 10x

If I am right that I am able to flourish and do things that would be unimaginable without AI, are there unintended consequences?

I am happy to produce content – own content, not AI generated – but where AI helps me with the distribution on different platforms, creation of shorts and videos, website creations and so on. But that also means an explosion of content, even if of good quality and authenticity. Ok, I am biased with my own content :).

In this “attention economy” we live today so it is not only poor AI generated content that will create noise, but we may also see fantastic new and authentic and human content that can be distributed more easily across different platforms and produced much quicker (tools like Riverside for podcasts as a concrete example).

This can mean an issue for platforms like this one, Substack, as this article so well tries to cover: The Real Reason that Substack is Collapsing. In short, we all just have 24 hours at our disposal, the same amount as Beyonce.

The danger of AI produced content is an old phenomenon and well known, but I am arguing about the abundance we may see also of actual good and human content, simply because of AI tools that make the legwork such as distribution and production so much easier.

Another obvious consequence is of course, if a 64-year-old retired ex-executive can start to create websites, patents, make lawyers not needed just to name a few examples, of course the impact on some kinds of professions are enormous. I honestly think that in some areas it will be a bloodbath (law, marketing, consultancy to name a few). But can we justify jobs that are no longer needed?

I would like to leave here with questions on these three areas and of course an invitation to watch the full video!

As a final thought, there are indeed a lot of links in this post. But as Andrew Sullivan once wrote in his classic essay Why I Blog, in The Atlantic:

“He is - more than any writer of the past - a node among other nodes, connected but unfinished without the links and the comments and the track-backs.”

Looking forward to your comments, as a post without a comment is a dead end. And of course, to having you join us on the journey towards the Azores!

So what are your thoughts and above all your personal experience with AI?

This is not an article saying that the changes in communication technology caused the changes, but they have often been an important tool for radical societal and political change.

I will end this essay to discuss the current state and the changes we need to be aware of.

The printing press – Reformation, enlightenment, and war.

15^th to 17th centuries

The invention of the printing press around 1440 by Johannes Gutenberg drastically reduced the cost of spreading text, thereby widening the spread of books outside the monopoly of the church. It meant the spread of new ideas and paved the way for the scientific revolution that for instance moved our word view from a geocentric perspective to a heliocentric view. It also meant a standardization of spoken languages into written languages, and the decline of Latin as a lingua franca started. The pressure to change the church started and lead to the reformation with thinkers like Luther and Calvin.

All these changes also lead to religious wars and conflicts not the least being the 30^th year war. It also led to power and financial resources moving from church to the states and private enterprises. The printed books and pamphlets also meant that stories of travel meant the start of colonialism from discovery into changing many European states into empires.

Newspapers, pamphlets and coffeeshops – The American and French revolutions

18^th to mid-19^th centuries

The spread of printed words gradually got less costly at same time as the beginning of urbanization meant that more and more people could interact on a daily basis. At the same time nothing much had happened to the speed of communications which was especially important in the American Revolution where grievances could be published, spread, and discussed in coffees shops that rapidly radicalized the population, while the government in London could takes months to react. The literacy had increased while the cost of printing had rapidly decreased, meaning that communication in cities could take hours instead of weeks or days.

The same things happened in France a few decades later partly being caused by the cost the French had spent supporting the American revolution, but also the pamphlets and founding documents being translated into French had a big impact. Key figures such as Marat in the revolution were publishers and journalists.

When the new republics consolidated it led to new wars such as the Napoleonic wars and the war of 1812 meant that wars spread to an almost global scale, was this really WW0?

During this time new legislations regarding the freedom of the press also came into place with the first country establishing one actually being Sweden in Sweden 1766. The limitiations on free speech and free publications also started from this time.

Telegraph, railways – Crimean war, American Civil war

Mid-19^th century – Early 20^th century

The telegraph and railways where being rapidly expanded starting from the Mid-19^th century meaning that information could spread almost instantly, and people could move in hours or days instead of weeks and months. The power of the newspapers grew with the introduction of large corporate media. The first war utilizing telegraph was the Crimean war both for the command structure and the transfer of news from the battlefront to the readers in London in days. Newspapers could at the same time be printed and distributed daily so the same news also reached other cities around Great Britain.

The speed in transfer of information increased the power of diplomacy which both led to both more rapid de-escalations and escalations.

The period also saw a rapid increase in literacy so that by the end of this period a majority of the population could also read. This could also pave the way for the creation of nationalistic movements that unified both Germany and Italy, which fundamentally changed the power balance of Europe, culminating in in World War I.

Broadcast Radio and Cinema – World War II

1920-1945

The early century saw two major inventions meaning a sender and receiver in closer proximity. The broadcast radio and the movie industries formed. Movies could have direct political consequences. For instance, the silent movie “The birth of a nation” had a direct connection to the resurgence of the Ku Klux Klan, and once the Nazis established a direct control of the movie industry it was used to indoctrinate the German population in violent antisemitism. During the war Hollywood also became part of the wartime propaganda machinery. An amusing example is Disney’s “The Fuehrer’s face” made in 1943.

Broadcast radio also became popularized. Radio receivers became more popular and established a more direct connection for leaders to enter the homes of the audience. In Germany the already established public broadcasting organization was organized under the ministry of propaganda, purging the organization of political opponents and jews. The government developed special low-cost receivers and during the war it became prohibited to use them to receive foreign broadcast from for instance BBC who took the unique position to not only send propaganda, but also transmit bad news which established a high level of trust in all actors.

Television and audiocassettes – The cold war and Iranian revolution

1945-ca 1990

During this period, we saw the expansion of the broadcast TV and during the later period Cable TV. Different countries had different versions of this mainly when it came to financing. The world saw the first global events both in terms of sports but also the space-programs. News became instant and fast while broadcast radio still had a stronghold. Broadcast radio became a tool to transmit propaganda with Voice of America and Radio Free Europe. News become a commercial product. In closed countries with censorship, it was difficult to reach the population, and an interesting example is how the messages before the Iranian revolution came through Audiocassettes.

The internet and social networks – The Arabian Spring

1990-2020

The rapid expansion of internet starting with email and use-groups made it possible to meet and discuss on a global scale which only accelerated with the creation of world-wide-web. During the early stages this was mostly a big open forum without other restrictions than potential moderation, but gradually this became more and more controlled with the introduction of social media. The importance of this possibility for everyone to become a creator became more competitive and controlled with the introduction of revenue share, making it possible to for influencer to make a living from their content creation.

During the Arabian Spring social platforms such as Twitter and Facebook were used to both spread ideas and to mobilize the participants. Facebook was accused of being an important part to spread information contributing to the Rohingya Genocide in Myanmar. The strength and the difficulty of moderating social media channels has led to both blocking in for instance Russia, Iran, and China, but it is still an important channel to mobilize activists from all part of the political spectrum. Of particular interest here are the algorithms which are normally set to maximize ad-revenues also has the side-effect of causing rage and polarization.

Another example of social media is the growth of ISIS and how they managed to radicalize young Muslims around the world through well produced propaganda.

AI and polarization

2020 – Now

The rapid growth of AI utilizing all the previous technologies with the side-effect that creation part is now more and more done by machines. What this will lead to is still unclear, but it could both lead to an environment of more truthful and less biased communication, but it could also lead to even more and better lies being produced. What the introduction of AI agents acting on social media will lead to is still unknown. Will they create the next set of revolutions in society? Or are we actually in the midst of such changes?

And maybe most important, how can we make sure that all the new technologies will benefit humanity.

Summary

There are of course many more examples of the connections between communication innovation and political change. We are probably in the midst of one such change just as we speak. I want to empathize again that the technology itself is not the cause and each of the changes has also led to positive changes. Also, from history we can see that even though the communication is global any harm or benefit is done on the local level in a human-to-human interaction. Also remember that any new communication technology comes on top of the already existing technology. We still have books and newspapers even though we get the newsflashes on social media we will search for news in the classical media. With this essay I wanted to put emphasis on these changes to make us aware of what it can lead to.

I think it is important to point out as well that the effect of these type of a novel communication technology has to be built on an existing set of real or perceived grievances mobilizing people from communicators into agents for change.

Today’s post is going to be a bit different. I just returned from Qingdao, the HQ of Haier, where they hold their annual innovation conference and awards ceremony. As a dean of a business school and professor of innovation working with a South African partner who works closely with Haier on “collaborative innovation”, I was very pleasantly surprised to receive an invitation to their annual Zero Distance Experience (ZDX) conference. Qingdao was a new city to me, though its beer, Tsingtao, was not.

At ZDX, I rediscovered a company I had studied over a decade ago: the Haier group and its unique business model, which is in constant evolution. Now, they are trying to capture the essence of their business model and distill it into a set of management principles that any business could and should adopt. However, when they start talking about the transition to the AI era, and about their vision for becoming AI-native, my skeptic’s radar goes off and I wonder if they have really thought this through or understand the implications of being AI-native.

During a keynote speech on September 21, 2026, Chairman and CEO Zhou Yunjie outlined his vision of AI-native organizations. As you can see from the slide above, in AI-First organizations “people still define the underlying architecture”; in AI-Native ones, people are no longer mentioned at all. To be fair, saying that agents “participate” in workflows implies that someone else does too, so this may be a matter of emphasis rather than exclusion. But I don’t think the omission is an oversight: in this vision, people look optional rather than integral. For a company that has been so forward thinking about unlocking human intelligence on so many levels, I’m not sure that their new vision has made room for human intelligence going forward.

Interestingly, before he shared his AI vision, he shared another slide in which he stated that “the central tension is the mismatch between the pace of technological progress and the pace of organizational transformation”. I don’t think anyone denies this, but he somehow seems to believe that this gap can be bridged, by Haier at least, while I see it as growing quickly into an abyss. People, and thus organizations, change incrementally at best, while AI capabilities are advancing at a pace few institutions can match.

Haier is a company I have read about for the last 15 years on and off within the context of new ways to organize with less hierarchy and more autonomy (some of you may be familiar with literature on teal organizations, the agile manifesto, holacracy, …). Haier has quite the unique business model, on the one hand, and, on the other hand, a constantly evolving way of explaining it. It is an ongoing test-and-learn experiment that started in 1984, when Zhang Ruimin took over a failing Qingdao refrigerator factory. I could summarize it briefly and a bit brazenly as rethinking companies as a team of teams, networked semi-autonomous micro-enterprises, focused on creating value both for the customer and the company.

It is hard to argue with success, especially when it is global, and spans across geographies and industries. This very-successful-yet-difficult-to-replicate model has been distilled over time into different attempts to capture the practice with frameworks, models, principles and, now, a theory. Practice came first; the theory is being (re)written afterwards.

Unlocking human intelligence

Haier has evolved its model many times. Here is my take on seven steps in its journey (dates are approximate, and Haier’s own version numbering has shifted over time):

Step 1 “OEC” (mid-1980s): Overall Every Control and Clear, a daily accountability system that tied pay directly to daily quality results. A discipline system rather than an autonomy system, it taught every employee to measure and own their results. Unlocking individual accountability.

Step 2 “ZZJYT”: self-management teams pursuing company-defined goals with almost no hierarchy; Zhang Ruimin says he removed the whole middle management level. The teams, later renamed micro-enterprises (MEs), typically have 10 to 15 members. Unlocking team intelligence.

Step 3 “RenDanHeYi” (proposed in 2005): “employees and users as one.” The user became the true boss of every team, and ME bonuses were tied to market results. In 2012, Haier restructured into some 4,000 micro-enterprises that hired their own staff, shared their own profits and set their own direction. Unlocking user-oriented innovation.

Step 4 “Zero Distance”: closing the gap between front-line employees and customers. Factory workers could see who had ordered the product they were making, and customer feedback fed innovation, moving Haier from mass production towards mass customization.

Step 5 “Ecosystem Micro-Communities (EMCs)” (2019): configurations of MEs around user scenarios, coordinated by EMC contracts, first on paper and later digital, that set roles, incentives and profit-sharing. An EMC can combine internal MEs, external partners and users. Unlocking ecosystem intelligence.

Step 6 “RenDanHeYi 2.0” (or 3.0: the London Business School case by Jacobides and Duke and others number it differently): digital contracting pushed the model beyond the firm’s borders, making the boundary between the institution and the “exstitution” (partners and users outside the firm) more porous. The logic travelled with acquisitions, where Haier asked local management to implement the model rather than sending in its own leaders.

Step 7 “Theory R”: Zhang Ruimin’s latest codification, in a book co-authored with Stuart Crainer of Thinkers50, Theory R: Rethinking How People and Organizations Work, due later in 2026. It seems designed to make a highly path-dependent model easier for others to adopt selectively. The results outside China are encouraging: at GE Appliances in Kentucky, according to Zohar (2022), profits grew at double digits within 18 months of the acquisition, and profit-linked bonuses were extended from 170 executives to nearly 3,000 employees. Bayer has built its own model, Dynamic Shared Ownership, which shares many of RDHY’s ideas.

While many aspects of RDHY should be familiar to those of you who follow business trends like agile, customer intimacy, teams of teams, etc. there is still much to like in the Haier approach and much to learn from their successes and, barring insight into their failures, into their unanswered questions (like how to maintain strategic coherence if they are truly decentralized).

They don’t speak about it in these terms, but to me the genius of Haier was unlocking, step by step, successive layers of intelligence: the individual, the team, the user and the ecosystem. But now this is where things will get complicated for Haier.

Managing intelligence in the age of AI

Until recently, managing intelligence in an organization equated to managing the people in the organization. AI is accelerating the move of a growing share of an organization’s collective intelligence from its people to its datasets, LLMs and models. So, if today most of the intelligence is still on the human side of the divide, in a short time more and more of the organizational intelligence will be artificial. And this will continue to accelerate. So the natural question is how to ensure, and do we want/need to ensure, that human intelligence remains the key factor in organizational intelligence? Today the answer seems to be a resounding yes. But as companies start to think of how to go from AI-enhanced to AI-native, it does not look like there is much room left for human intelligence in the intelligence stack.

And this is where the Haier story gets murky. Founder Zhang Ruimin, who handed over the chairmanship in 2021, has always had a strongly human-centric approach to business. His successor as Chairman and CEO, Zhou Yunjie, has gone as far as making RenDanHeYi one of the group’s key board committees. Yet at ZDX the group’s sights seemed clearly set on an AI-native future. Where do people fit in here? Except for validating decisions and recommendations made by AI, it is hard to see. Interestingly, researchers have already begun describing Haier’s EMCs as AI-driven digital business ecosystems. The unanswered question is what RenDanHeYi’s own logic implies when some/most of the intelligence at the edge is no longer human.

Not just a Haier problem

This problem of finding a role for human intelligence in an AI-native reality is not just Haier’s challenge for the years ahead. I think we are collectively naïve about this future with many people reassuring themselves that we will always need human taste, critical thinking and decision sign-off - even in an AI-driven world. Allow me to suggest that today when you ask AI “what should be our next innovation” it can do a lot before it makes well-argued recommendations. But we are rapidly approaching the point where, for power users, if your AI is agentic, it can design, research, prototype, test, deploy, create commercial websites, gather pre-orders and data, run synthetic and real-world focus groups, and come back with actionable stress-tested recommendations that if you put product X on the market you should be able to generate Y revenues at Z profitability. If you credential the AI, the product could even be launched and generating revenue before you sign off on it.

If you think this sounds like science fiction, consider this: according to its own case study, Board of Innovation built an AI-native innovation system for a leading consumer-goods company in which up to 400 AI agents operate in parallel to turn proprietary and real-time market data into consumer insights, product concepts and synthetic validation (testing concepts against AI personas rather than real customers).

To be fair, the most rigorous counterargument comes from Princeton’s Arvind Narayanan and Sayash Kapoor. Their “AI as Normal Technology” framework holds that AI is transformative, but that its impact is paced by diffusion rather than by capability: organizations, regulation and complementary investments slow adoption, as they did for electricity and the internet. In a June 2026 essay, they describe software work as a “decide-execute-deliver sandwich”: deciding what to build (framing, specification, planning), executing it (design and implementation) and delivering it (testing, verification, integration, maintenance). AI, they argue, “has compressed the middle of the sandwich, but has left the two ends largely unchanged.” In his ICML 2026 keynote, Narayanan argued that we should be building precisely the complementary skills, such as judgment and taste, that are necessary for the decide and deliver layers. While I do gain hope from this view, and not-so-secretly wish it to be a viable scenario for the future, my instinct is to call this out as naïve for similar reasons as for Haier.

Thinking that, as AI ramps up its abilities in nearly all domains, we will somehow be able to understand the big picture, all the data that the model crunched, and the swarm of AI agents deployed well enough to make informed decisions seems difficult to imagine. In a world in which it is almost impossible to understand the recommendations made by AI systems at the necessary granularity, it seems like “go ahead” will be the default decision. In contrast to Narayanan and Kapoor, I can see the deliver layer becoming increasingly AI-led (think of early experiments with digital twins and synthetic control arms in clinical trials), with the decide layer following quite quickly (think of the ways AI is already accelerating biotech discovery). In academia, I am struck by how much the quality of research that AI can produce end to end, from framing to write-up, has improved in just the last six months.

That said, every organization with investors will still need a fall guy. And that person will be the manager tasked with validating AI recommendations and decisions. When things go sideways because of AI, the responsible manager will get fired. The pattern is already familiar from ordinary software: Sonos’s CEO stepped down in January 2025 after a botched app update alienated customers. That was not clearly established as an AI failure, but it shows where accountability lands. So, in a short few years, will accountability and responsibility remain the last bastion of human contribution to the organizational collective? I’m guessing that it will. This is, in effect, Narayanan and Kapoor’s “deliver” layer; where I differ is on whether it will remain human in substance or only in signature. And that is not nothing: accountability is a genuinely human function, rooted in trust, responsibility, morality, law and governance. The open question is whether it is a foundation on which to rebuild human work/contribution/value, or merely a rubber stamp that designates a scapegoat every time an AI fumbles.

And this leads us to the increasingly obvious and uncomfortable question of whether or not a role exists for humans in an AI-native business environment. I don’t know the answer, but I can draw a parallel. Just like dark factories that only have a few token humans there to make sure the robots are behaving as planned, there may not be much need for humans in dark businesses. That is my prediction. I asked Zhang Ruimin about this in Qingdao during a private conversation over dinner. He simply stated that he was still thinking about this. But what he did know is that if a dark factory is producing products consumers don’t want, it won’t last long. That answer toes the line of RenDanHeYi: value is defined by the user. If understanding what users truly want remains a human advantage, that may be where human intelligence keeps its edge. But, the day AI comes to know users better than we do, even that bastion falls.

I do hope the “dark business model” will not emerge as a dominant business practice. We will need to check in again in five years to see if the AI-native companies that thrive are those that have redesigned business models around human judgment rather than removed it.

These were the musings of a 4D: a Dreamer who is also a Doer who is trying to think proactively about how to Design organizations for the AI age but Doubts our ability to see clearly into this nebulous future.

Sources

Board of Innovation. (2025, November 17). Scaling autonomous innovation in the enterprise with 400 AI agents in parallel [Case study]. https://www.boardofinnovation.com/cases/scaling-autonomous-innovation-in-the-enterprise/

China Daily. (2021, November 5). Haier’s Zhang stands down as chairman. http://qingdao.chinadaily.com.cn/gfqd/2022-09/27/c_815995.htm

CNBC. (2025, January 13). Sonos CEO Patrick Spence steps down after app update debacle. https://www.cnbc.com/2025/01/13/sonos-ceo-patrick-spence-steps-down-after-app-update-debacle.html

Crainer, S. (2022, January 20). To boldly go. Thinkers50. https://thinkers50.com/blog/to-boldly-go/

Fischer, B., Lago, U., & Liu, F. (2015, April 27). The Haier road to growth. strategy+business. https://www.strategy-business.com/article/00323

Haier Group. (n.d.). Corporate culture. Retrieved September 30, 2026, from https://www.haier.com/global/about-haier/culture/

Jacobides, M. G., & Duke, L. (2020). Haier’s (2019) ecosystem revolution: From Rendanheyi 2.0 to Rendanheyi 3.0 [Case study]. London Business School. https://publishing.london.edu/cases/the-haier-cases-c/

Krivitsky, A. (2025, February 4). Studying org designs of Haier’s RDHY and Bayer’s DSO. Org Topologies. https://www.orgtopologies.com/post/studying-org-designs-of-haier-s-rdhy-and-bayer-s-dso

Li, H. (2023, April 26). A business of its times: Haier’s self-evolving organization [Interview]. McKinsey & Company. https://www.mckinsey.com/capabilities/people-and-organizational-performance/our-insights/a-business-of-its-times-haiers-self-evolving-organization

Minnaar, J. (2022, September 14). The EMC contract as a smart coordination mechanism. Global Focus. https://globalfocusmagazine.com/the-emc-contract-as-a-smart-coordination-mechanism/

Narayanan, A. (2026, July 13). What will be left for us to work on? [Keynote address, ICML 2026]. AI as Normal Technology.

Narayanan, A., & Kapoor, S. (2025). AI as normal technology: An alternative to the vision of AI as a potential superintelligence. Knight First Amendment Institute at Columbia University. https://kfai-documents.s3.amazonaws.com/documents/2b27e794d6/AI-as-Normal-Technology---Narayanan---Kapoor.pdf

Narayanan, A., & Kapoor, S. (2026, June 11). Why AI hasn’t replaced software engineers. AI as Normal Technology. https://kfai-documents.s3.amazonaws.com/documents/2b27e794d6/AI-as-Normal-Technology---Narayanan---Kapoor.pdf

Steiber, A. (2023). Ecosystem brand and strategy in the new economy [Article]. LinkedIn. https://www.linkedin.com/pulse/ecosystem-brand-strategy-new-economy-dr-annika-steiber

Steiber, A., & Alvarez, D. (2024). AI-driven digital business ecosystems: A study of Haier’s EMCs. European Journal of Innovation Management, 28(8), 3966–3984. https://doi.org/10.1108/EJIM-01-2024-0076

Thinkers50. (2026, August 31). The road to Theory R. https://thinkers50.com/blog/the-road-to-theory-r/

Zhai, S., & Jia, D. (2021, November 6). Haier founder Zhang Ruimin to step down as chairman. Caixin Global. https://www.caixinglobal.com/2021-11-06/haier-founder-zhang-ruimin-to-step-down-as-chairman-101801192.html

Zhang, R. (2019, February 3). Haier’s philosophy: Personal values first. Global Focus. https://globalfocusmagazine.com/haiers-philosophy-personal-values-first/

Zhou, Y. (2026, September 21). [Keynote address on AI-native organizations] [Conference presentation]. Zero Distance Experience (ZDX) conference, Qingdao, China. Photographs by the author.

Zohar, D. (2022). GE Appliances: An American catalyst. In Zero distance: Management in the quantum age (pp. 195–201). Palgrave Macmillan. https://doi.org/10.1007/978-981-16-7849-3_18

What if focusing on problems just creates more problems and keeps us from finding solutions not just independently but in our existing systems?

In this conversation I talked with Ingela Svedin on why our 20th-century ideas are breaking down and how we can transform our institutions into living systems. Ingela has had a long career in places like the Swedish Institute and the Swedish International Development Agency, as well as leading Inner Development Goal work at Klustret Ekskäret. She tied together a host of ideas and movements (links below) that changed the way I see the big picture today.

Explore references from conversation:

• Nora Bateson, International Bateson Institute: Warm Data

• Planetary Boundaries

• Kate Raworth and Doughnut Economics

• Hospicing Modernity: Facing Humanity’s Wrongs and the Implications for Social Activism : Machado de Oliveira, Vanessa

• Tomas Björkman

• Yuval Noah Harari

• Theory U | Leading from the emerging future - Otto Scharmer

• Inner Development Goals (IDGs)

• Quintet of change

• Loud Summer

On the 8th, 9th and 10th of October 2027, the 4D community is going to the Azores. Nine islands in the middle of the Atlantic, closer to nothing than almost anywhere in Europe. Ida found the place, which is less of a coincidence than it sounds since she now lives there and is our woman on the ground. Heraldo has been holding the thread together since three in the morning for weeks. The dates are real. We have a hotel, we have a plan, and we have a year to prepare it.

It will not be a conference. There will be no session tracks, no panels and no slides. We will walk, eat together, and have the kind of long, unhurried conversations that are almost impossible to have anywhere else. That framing is not ours. Björn wrote a piece here at the start of August about what a modern pilgrim is actually looking for, and he used a phrase that has stayed with all of us since: solvitur ambulando, it is solved by walking. Ida sharpened it further in her latest 4D conversation. A pilgrimage, she said, is really a journey towards ourselves.

Here is the part I want to be clearest about, because it is the thing most likely to be misheard. The trip is not the product. The year is. If we all fly to a beautiful island and have three good days, we will have had three good days. What we are proposing instead is that we spend the next year actually learning something together, and that the Azores is where that lands. The island is the fertile ground for our discussions, a milestone in the journey (not the offer).

When I presented this to the community, I used a three horizons approach, because it helps us think ahead to the longer-term impact. In eight weeks, what do we learn? In one year, what can we build? In ten years, what did we achieve? Notice the tense moving from present to conditional to past, because that is the whole idea. And notice the word “we”, which is the harder shift. Henry Mintzberg said something to Nathan on this Substack back in May that I have not been able to put down: the “I” question has to have a “we” answer. If there is a thesis to this entire journey, that is it.

What will be our journey?

The vehicle is the Slow Schoolhouse. I wrote about it here in March, and the short version is this: small circles of four to six people, each taking one topic, meeting over a few weeks with a rotating steward rather than a permanent expert, and finishing with something concrete rather than just a good discussion. Nobody sits at the front of the room. The person stewarding a session is simply a fellow traveller who prepared a little more that week and hands the baton on at the end.

We propose initially eight topics to open:

• AI and how we think.

• Institutions and power.

• Democracy and trust.

• Truth and authorship.

• Meaning and purpose.

• Attention and time.

• The planet and the systems we live in.

• The next generation, and what we are handing over to them.

How many of those actually run, and how far each one goes, is something only our commitment as community members can decide. It depends on the effort we put in, individually and collectively. A topic only becomes a circle when someone says they will help prepare it, and a circle only goes deep if the people in it show up week after week. Eight is what we are offering. What we end up with is what we build.

Each circle will suggest a book or an essay or a talk worth arguing with, but let me be direct about this because it has come up more than once in our WhatsApp groups: the book is the catalyst, not the homework. Read the whole thing, read a chapter, listen to a podcast, or bring nothing but your own experience. The commitment is not to read the same pages. It is to turn up ready for a serious conversation about the same question.

Every circle should end with three things: some curated knowledge to share with the rest of the community, a proposal for a 4D conversation, and an idea for a Little Big. Something small enough to actually do and meaningful enough to matter. We have two reference cases in our own archive already. Henrik Pålsson and what connectivity and renewable energy did for a struggling island in the Philippines, and Vergis in India, whose work in education is still the thing I go back to when I need to remember why any of this is worth the effort. Neither started big. Both compounded. That is the shape we are looking for.

Collective knowledge is what a circle produces. Collective agency is what it is for.

There is a reason for those three outputs, and it goes back to something I wrote here in March. We do not have a content problem. We have more good material available to us than any generation in history, and it is not making us more capable of acting on any of it. What is eroding is collective efficacy, Bandura’s term for a group’s shared belief in its own capacity to do something together. It disappears quietly, through the daily experience of being hyper-aware of problems and completely unable to affect them. Curated knowledge alone does not rebuild it. Knowledge, plus people you trust to think with, plus something you actually do at the end of it, might. Collective knowledge is what a circle produces. Collective agency is what it is for.

Then, in October 2027, we arrive somewhere we have never been, having already spent a year thinking about the same difficult questions. We walk in small groups. We disagree on trails and continue over dinner. There will be quiet time, and there will be an alternative to walking for anyone who wants one. And on the last half day, we ask what comes next. The point is that we start those conversations somewhere interesting instead of at the beginning, because we will already share a language for them.

Now the honest part, because a Doubter should never be the last person to speak. We are aiming for twenty to twenty-five people, not two hundred, and that is a deliberate choice for depth over scale. Several things are genuinely undecided: what it costs, when a commitment becomes binding, and whether we as a community are ready to run circles ourselves rather than have one person organise everything. That last one matters most. One idea being discussed is that over time more of our 4D Conversations would be hosted by the Slow Schoolhouse teams themselves, rather than always by the same few people. It is a proposal, not a decision, and it should come from the circles rather than be handed to them. Tell us what you think.

We are not starting from zero. Since January this community has published more than twenty pieces, from eight different writers, alongside conversations with Mintzberg, John Hagel, Thomas Björkman and many others. What has been missing is not content. It is the move from talking to building. On the 23rd of September we are holding the first step towards the Azores, an open session where we will share the topics, name the first circles, and answer the questions we have not answered here. If something in this piece named a feeling you recognise, that is the room to be in.

This post is coming to an end, but the story is only beggining. And if you are reading it, thank you for helping to co-author the journey.

Casi Juanes writes here about education, critical thinking and AI. He is one of the 4D community’s Substack writers.

Photographs from the Azores by Ida Helena Rust.

At the end of the conversation, Casimiro Juanes also introduces our journey towards the 4D Community Gathering in the Azores in October 2027 - and a simple way of thinking about the road ahead: 8 weeks - What do I learn? One year - What can I build? Ten years - What did I achieve?

A fitting invitation to turn reflection into movement. Solvitur Ambulando.

I had prepared another post on AI’s impact on the world, when the world lost what can only be described as an amazing woman.

Dolly Parton had always seemed larger than life to me, in the way any band from the 1970s/80s seemed larger than life. With her big hair and personality, she seemed to be thrust into a box by the music industry. While she embraced it for a while – she also had the courage to step out of it and became more.

Yes, she always championed her feminism (read looks) to move her ahead, but she also set out to do more for literacy in the United States. Her dad had never learned to read or write and she wanted to make sure that her nonprofit raised both literacy and education rates in places that were severely lacking and the employees at her theme parks were given the opportunity to go to college through scholarships. It’s rumored that a number of high school (A-Level equivalent) bands had their uniforms paid for by Dolly Parton. Parton has been instrumental in funding for hospitals, covid vaccines, disaster relief, animal welfare and conservation. In short, Dolly Parton was a woman who worked hard, earned well and donated her time, money, and name just as equally when it was deserved.

So, why bring this up? Because quite frankly she was very specific about her wish to not become an eternal digital presence, or AI clone. In 2023 she expressed concern about the use of artificial intelligence with quotes like:

“I think I’ve left a great body of work behind.”

“I don’t want AI. The only intelligence I have is artificial,” the musician joked. “Everything I have is artificial.”

^{Quote sources: Source: Guitar.com, Independent.com}

This is something important to think about – especially since it was rumored earlier this month that she was or her team was exploring Avatar performances _{(Source: Billboard)}

In 1997 Fred Astaire appeared in a Dirt Devil commercial – dancing with a vacuum. Astaire who was known for dancing with coat racks, mops and light poles in his movie musicals (Easter Parade, Royal Wedding and Dancing the Rain) had his film meshed via computer generation – which caused quite a stir within the Astaire family itself – since not all members agreed it should have been done. (Source: Variety)

As a fan of movie musicals – I was confounded and disappointed by the commercial in 1997. I was also simultaneously impressed by what CGI (computer generated images) were able to do at the time. But, it wasn’t the first time I’d seen it. Gene Kelly danced with Jerry Mouse (of Tom and Jerry) in Anchors Aweigh in 1945, so it wasn’t all that impressive. But as I wrote in my college essay was that it as portent to what was to come…

Fast forward and we have entire movies being created with CGI that include virtual “actors” ((need I mention Jar-Jar?). Now as AI becomes even better and the world lives its life in digital we have to worry about deep fakes, AI generated images and more.

No longer is it enough to photo-chop* your face onto Dolly Parton’s body. Now AI can mold your features to hers and produce an image that is downright scary. Which brings this all around to what I was writing about last time. The use of AI images. And my comment about “image-based-abuse.” [Again I refer you to Janhavi Kapoor’s comments on the topic.] In Run, Rose, Run, a children’s book by Dolly Parton it says “When a girl didn’t have anyone to stand up for her, that girl had to stand up for her own damn self.” And while many will think I’m being sexist and I probably am in a small way, woman and celebrities are going to be more in the line of fire for AI regeneration than anyone else, both while alive, but even more so when dead.

Parton’s death returns me to the question of what to do when my mother is gone. Not only do we need to set-up the provisions for her estate, but we also need to get a clause added that expresses the wishes of the use of digital materials. As of 2026 there are not set laws that restrict the use of image an voice after death. You have to outline what can and cannot be done with your likeness and vocality just like you have to outline who gets the horrid green lamp from the living room or the cat. I highly recommend reading “AI and the Digital Afterlife” from the Grimaldi Yeung Law Group.

Tennessee, home of Nashville and country music, actually passed into law the Tennessee ELVIS Act in 2024 that expanded protection for voice and likeness – but only in Tennessee so Parton is covered). There are other states following suit, and the EU is making moves to put some provisions in place as noted by Jones Day Law Firm. But will it be enough. After all, once you’re dead and gone, you have no say over what happens. Your can only hope your wishes will be upheld.

So, with the passing of a “rhinestone life that shone bright enough for the world to see” Dolly Parton’s legacy will live on – but it will unlikely to be in AI-form. At least we can hope… And with the rumor before her passing, but her comments made in 2023, the future of Dolly’s digital presence is uncertain – at least to the public at this time.

And so I leave you with this Dolly Parton quote: “Don’t get so busy making a living that you forget to make a life.”

Some other reading:

• Proskauer Law Firm, June 2024, The King is Back (in the Digital Era) | The ELVIS Act, Generative AI and Right of Publicity

• Lexology, October 2024, Legal Rights of the Undead: The Spooky Implications of Legacy AI

• Slashdot August 2025, The Dead Need Right To Delete Their Data So They Can’t Be AI-ified, Lawyer Says

Many thanks to my colleague Björn Rudberg for the use of his photo.

*Photo-chop - I deliberately chose this term rather than name a specific photo editing software.

Note: None of this was authored by AI, unless you count the brAIn.

A shadow looms over the Earth. Random fluorescent and flickering red dots break through the gray mass. Only in a remote corner of the Earth is a bright, warm, yellowish light visible. Gloomy industrial landscapes everywhere. What was once a blossoming tree is now a dry trunk, a sculpture from bygone times. Closer to the ground, the activity of countless entities is visible. Like an ant colony, which, seen from space, forms perfect geometric drawings.

Figures resembling humans stare ahead with hollow eyes. They do not speak, yet seem to communicate and coordinate their actions flawlessly. Like machines, they are attuned to one another. Sometimes they seem to shiver. As if thunder rolls through them. As if something is watching them. Inside them. There is no escape. Even their movements are not voluntary. A bionic arm raises itself to solder a metal plate onto a machine; a head tilts slightly for perfect vision. A centaur-like robot-human creature spans distances the size of a stadium in just one leap.

Somewhere in this grey, industrial world, a Light shines. It is the brightest Light ever conceived, as if all energy is drawn toward it. Lush trees and flowers everywhere. Perfect-looking children play merrily with the animals: majestic lions, black panthers, cute lambs, colorful parrots, jolly barking golden retrievers, you name it. There is no violence, and all play together. Human beings talk animated and have the most radiant skin, their faces are perfectly symmetrical, their hair gleaming, and their bodies like those of Roman gods. Only the word ‘divine’ can describe the Light.

Defining ASI
ASI is often described as a single, super-powerful and intelligent AI. One, whose “intellect greatly outperforms the best current human minds across many very general cognitive domains” – as defined by Nick Bostrom. Hypothetically, this could also be a biological intelligence as the result of genetic engineering. Roman Yampolskiy defines ASI as “a computational agent that outperforms all other intelligent agents over all possible environments”. This idea is based on the fact that our intelligence has given us an evolutionary advantage.

An alternative perspective to the Singleton ASI is Pedro Domingos’ Master Algorithm. This is a universal learning algorithm sitting on top of a hierarchy of algorithms, consisting of controllers and commanders. The latter create new algorithms for system optimization. While Domingos does not specifically identify this as an ASI, the concept nevertheless reflects the idea of a superior algorithm capable of controlling and optimizing the world by means of algorithms.

Because there is no certainty whether ASI will be created, I would like to propose a different understanding of ASI, namely: ASI as a ubiquitous system of distributed and decentralized cyber-physical-human systems…
I know, this is quite a mouthful, and I will break it down in a moment. A much simpler way to describe this ASI is to imagine the Internet of Things (IoT) everywhere, with AI running through it. This ASI does not require the manifestation of a Singleton ASI, and the advantage is that we can start to experience what this ASI may be like right now.

IoT devices are connected devices. The device itself is physical, while its sensing, communication, and data processing take place in the digital realm. Hence: cyber-physical. When connected devices intentionally interact and operate together toward a common objective, they form a system. As there are increasingly more connected devices with a common goal, a great many systems are emerging.

Originally, connected things were dumb. They were programmed in a certain way and carried out that task, in most cases under the supervision of humans. Now, with the addition of AI added to these things, they are becoming increasingly self-adapting. These devices are able to determine for themselves what needs to be done to accomplish what they are supposed to do in the most efficient and effective way. This is why they are called smart: smart cities, smart homes, smart cows, smartwatches, etc.

In the early days of IoT, for example, cows were simply equipped with GPS tags that tracked their location and sent the data to a monitoring device. Today, these devices do much more, measuring a wide variety of metrics such as health status, methane emissions, and milk production. This data may no longer require a specialist or specific device to interpret and display it. Instead, it can be received on a tablet or smartphone and presented in such a way that a farmer can make sense of it without training. This expansion of what can be monitored and controlled is only growing.

Now imagine, closer to home and a bit further in the future, your refrigerator communicating with both the milk carton and your AI brain implant. The implant analyzes data from nanobots in your gut and lets the refrigerator know, within a fraction of a microsecond, that you need a different drink. The refrigerator then communicates with a smart manufacturing plant, which internally, based on a combination of this gut-derived data and healthcare information produces a personalized drink. The drink is delivered by an autonomous vehicle controlled by a logistics AI, which plans an optimal delivery strategy in coordination with other deliveries. Your nanobots detect whether you have actually consumed this health-boosting drink, and this information is shared with your health insurer, which is also entirely run by algorithms.

This does not require superintelligence, but simply the connection of different devices and sufficient computing capacity to process the data and control a multitude of devices. “Simply” is, of course, the result of highly complex, networked AI development. Still, relative to a Singleton-ASI, it is far more feasible.

What I described above can be viewed as a decentralized system. Here, each node has a certain degree of autonomy and determines its optimal calibration through interaction with the other nodes. In the case of a Singleton ASI, there would be no decentralized system. Instead, the ASI would be the main controller of a centralized system. A distributed system, on the other hand, the computational activity is spread across the various nodes, all of which work together on a predefined goal.

Translated into our human world, a centralized system is like an old-fashioned kingdom, a decentralized system is like the United Nations, and a distributed system is like a global logistics network.

ASI is the Network
Back to the ubiquitous ASI. This requires, again, a bit of imagination and abstract thinking. AI will be, and to a certain extent already is, able to create new algorithms, materials, and devices. In the future, there will be countless decentralized and distributed systems, all of which will interact with one another. They will collect and exchange immense amounts of data, which will then be processed to serve different objectives.

AI is computationally rational, which means that it optimizes the goals for which it has been trained. Human emotions such as jealousy or preferences based on affection are unknown to AI. Therefore, all these AIs will negotiate the most optimal settings for themselves and for the systems they serve, in order to maximize the achievement of their respective goals.

Now imagine, along with the futurist and inventor Ray Kurzweil, that everything will contain AI: the entire universe. Then this will effectively be a closed system: the ultimate system of systems. Due to its rational nature, this universal AI would evolve toward a self-maintained equilibrium. Having access to all available data, both past and present, it would initially reshape the future to achieve desired outcomes, ultimately gaining full control over space and time, and eliminating all uncertainty.

We, as biological human beings, will be assimilated into these systems, and consequently into ASI. We have become cyborgs; or perhaps entirely artificial beings. Even if some of our biological intelligence is preserved, we will also be fully computationally rational.

Nowadays, our emotions are seen as a unique characteristic that distinguishes us from AI. However, through affective computing technology, AI is increasingly able to recognize our emotional responses and anticipate them by either changing our environment or managing the emotions that are likely to occur. If a wide range of emotions persist in the future, they too will be subject to optimization. That is to say, they will persist only when they serve a purpose established by ASI.

Like the canary in the coal mine, our emotions may be used to signal unwanted deviations because ASI has determined that this is more economical or more accurate than using another type of device.

By integrating more and more AI into everything, we are actively building this future for ourselves. That is, we are building it only when we increasingly rely on AI judgments and disregard our own. Even if it means that our judgement is sometimes wrong, resulting in sub-optimal outcomes. If we dare to overrule AI, or at least engage in a critical dialogue with it to establish shared control, this deterministic ASI might well not rise after all.

I found myself asking this question again while watching Nolan’s The Odyssey with my cousins when I was in Jakarta last week, in between meetings with the financial regulator and some banks and crypto exchanges.

Odysseus returns home after twenty years, but Ithaca is no longer the place he left, and he is certainly no longer the man who left it. His wife Penelope does not recognise him. His son Telemachus does not recognise him. The people who once around him see only another tired traveler. Only Argus, the old dog, knows who has come home. There is something piqued my interest about that scene. No elaborate speech, no dramatic reunion. Just an old dog sniffing and lifting his head one last time, as if to say, There you are. I knew you would come back.

I don’t know why that scene got me. It just did.

One line stayed with me long after the film ended: “No one knows who I am anymore.” Maybe that’s what happens to us on long journeys. You stop being easy to recognise, even to yourself.

I left Sumba, the island on the Indian Ocean, more than thirty years ago to study in Jakarta. Jakarta eventually led me to Copenhagen, my first journey outside Indonesia as a professional for Ericsson and from there life took me through different parts of Europe, Asia and America, with short business trips to Africa. After living in the USA, eventually Sweden became home, and I have now lived here for a decade.

When I look back, I sometimes wonder which version of me is the real one.

The boy who grew up on the coast of Sumba, looking across the Indian Ocean towards Australia?

The teenager who left for Java to study, following in the footsteps of my grandmother, who in 1918 became the first person in her family to leave for a higher education, travelling on a Dutch cattle ship? I have often wondered what it must have taken for her to make that decision at a time when leaving the island was not something people simply did.

The young man who later guided, hiked and climbed mountains on four continents?

The business professional who arrived in Scandinavia from a strict Calvinist upbringing and experienced a fairly serious culture shock?

The person who lived in the United States and Southeast Asia before eventually making a life in Sweden?

Or all of them at once, badly stitched together?

Every place has left something behind. A language starts appearing in your thoughts without you noticing. A different culture teaches you what’s considered polite, or ambitious, or generous, or strange. I sometimes curse in Swedish, negotiate in English, apologise in Indonesian and mumbling in Wanukakan without deciding which version of myself should show up. Living between many cultures has made me realise identity is rarely a fixed object. It’s more like a coastline. Familiar in shape from a distance, but the ocean keeps taking something away and leaving something else behind.

I grew up on the coastal side of Sumba, beside a vast ocean and its current that never to come from only one direction. Tall waves overlapped one another, currents collided. Maybe that’s why the idea of identity as one single, stable thing has never made sense to me. We like to imagine that underneath all our experiences there’s one fixed self waiting to be found, but a human being might be more like an island surrounded by ocean than a solitary stone. We’re made from waves, currents, memories, places, relationships, and whatever else keeps moving through us.

When I was at Phaleron Bay in Athens, the Greeks had another way of asking the same question. The Ship of Theseus replaces every plank over time until none of the original wood remains. Is it still the same ship? It sounds like a puzzle for a classroom, but after enough countries, languages, friendships, losses and new beginnings, it stops being abstract. How much can change before you become someone else? And what is it, exactly, that lets you keep saying “I” when so much of the material around that “I” has already been swapped out?

This is also why transhumanism has never felt that alien to me. We usually talk about it as if it begins with artificial intelligence, or Brain-Machine-Interface neural implants, or machines becoming part of our bodies. I think it began a lot earlier than that. Humans have always used technology to get past the limits of the bodies and environments they were born into. Boats let us cross oceans. Writing let memory outlive the brain. And at some point somebody figured out how to make a machine that could think alongside us too.

When I think about my own life, I realize how much these technologies have already changed what kind of existence I get to have. My ancestors could spend an entire lifetime inside a fairly small geographical world. I’ve left an obscure island in Indonesia, built a life and worked across Asia, America, Europe continents, traveled through Africa, kept friendships alive on the other side of the world, and gone back to Indonesia in hours instead of months. I’ve had few hard winter in my rustic cabin in the Arctic Circle in Lappland, and the heat of the tropics in my grandparents house on the equator by the Indian Ocean, and moved between cultures with completely different ideas about time and relationships. Most of my working life now fits in something I carry in my pocket. None of that would have been possible for the generations before me. I don’t think of myself as less human because of it. I just live in a world my ancestors couldn’t have imagined.

Maybe that’s what’s happening again with AI.

I don’t see it as something suddenly foreign to being human. I see it as another technology I’m learning to use, the way earlier generations learned to use writing, engines, telephones, computers, the internet. The difference is that this one doesn’t just extend my hands or my eyes or my ability to travel. It’s starting to extend parts of my thinking. I can ask a machine to challenge an opinion, steelman an idea, translate something, help me understand a subject I know almost nothing about, or work through an argument that would once have taken hours in a library. That doesn’t mean the machine has become me. It means the edges of what I can do as one person have moved again. The same thing happened when someone first learned to read and started carrying the accumulated thoughts of people who’d been dead for centuries around inside their own head.

Even clothing, something so ordinary we barely think of it as technology, changed not just what humans could survive but who they could become. For a Sumbanese , an ikat (our traditional clothing) does more than cover a body. It signals family identity, belonging, status, history. We don’t just use technology. Over time it becomes part of how we recognise ourselves and each other.

My own work has made this question more complicated, not less. I spend a lot of my time thinking about data, privacy, and how organizations can collaborate without exposing the sensitive information that represents real people, in order to tackle shared problems and protect our societies. A person can be represented by a phone number, a transaction, a behavioral pattern. None of those things is the person. You can describe someone completely in data and still not know them at all. I see this constantly in my own work and it still gets to me. AI is going to make that distinction more important, not less. The better machines get at predicting us, the more it matters to remember that a prediction is not a person.

If an algorithm knows what I’m likely to buy, where I’m likely to travel, what I’m likely to read, maybe even what I’m likely to think - does that tell it who I am? Or does it just build another version of me, the way my passport does, or my job title, or my old photographs?

The Ship of Theseus again. If my body changes, my memory is partly stored outside my brain, my relationships live across digital networks, my thinking is augmented by AI and the cultures around me keep shaping how I act - what’s actually left?

I don’t know.

The older I get, the less I believe identity is something you build once and carry through life. It feels more like something that’s constantly being negotiated between what you inherit and what you run into along the way.

Maybe the self isn’t the original wood of the ship. Maybe it’s the continuity of the voyage.

And that brings me back to Argus.

He didn’t recognize Odysseus by his clothes, his face, his title, or the stories he could’ve told about where he’d been. He recognized something none of that could fully describe. Twenty years had changed Odysseus - apparently not enough to erase whatever Argus had known in him before he left.

Sometimes I wonder what Argus would make of me now. Not the teenager who left Sumba. Not the young man who surrendered to his fate and came out the other side of a snow blizzard at 6,000 meters on the China-Nepal border. Not the person who’s chased the best peaks and trails the American continent has to offer, from McKinley in Alaska to Cerro Torre in Chile. Not the person who eventually made a life in Sweden. Not even the version of me who spends his days thinking about technology and data. Just whatever’s still underneath all of that.

I like to think he’d pause for a second, look at me, and wag his tail anyway.

Maybe that’s what coming home actually means. Not returning to the place you left - that place has changed, and so have you. Something quieter. Discovering that after all the countries, the languages, the technologies, the identities you’ve picked up along the way, there’s still something that can be recognized.

Maybe that’s why Homer’s The Odyssey still holds up after thousands of years. It’s not really a story about a man trying to get home. It’s a story about a man who’s changed so much that almost no one can recognize him anymore, and yet somewhere, somehow, someone still does.

Technology has already changed us. It always has.

The question I keep circling is whether, after all of that, there’s still something in us an old dog could recognise.

Welcome back! Hope you all had a great Summer break.

We are starting really strong with Ida Helena Rust as our guest. Below the details:

When AI becomes a god, will you have faith? – 4D Conversation with Ida Helena Rust - August 27, 18:00–19:30 CET

We have an extraordinary 4D Conversation – Dreamers, Doers, Designers and Doubters – on August 27th, addressing possibly one of the biggest challenges of our times.

Ida Helena Rust has dedicated a decade to fusing her knowledge in Philosophy and Technology, specifically AI, as part of her PhD thesis:

Human Critical Thinking and Creativity and the Rise of the Artificial Superintelligence (ASI) God

She uncovers profound questions: Can we still think critically in the presence of the ASI-God? Where do we as humans end and machines begin? What does it mean to be human?

This is an unbelievably important conversation for anyone interested in – and concerned about – the development of AI.

All AI discussions are important, but here we will explore perhaps the most fundamental one: Are we threatened in our own humanness? Will we lose control over AI?

The date and time:
August 27th, 18:00–19:30 CET

Place:
Digital Event – send a mail to hello@4dcommunity.org and we will send you the link.

To build on this positive momentum...

We have the immense pleasure of sharing that we have certainly not stayed put during the summer 😊. As the saying goes, a crocodile that stands still eventually becomes a handbag.

1. We continue to grow in many positive ways on Substack. We are now 10 writers, with three of us focusing mainly on podcasts. If you are not already subscribing, what are you waiting for?

The 4D Community - Dreamers, Doers, Designers and Doubters | Heraldo Sales Cavalcante | Substack

2. We have launched our new homepage:

4D Community - Dreamers. Doers. Designers. Doubters.

3. We are now present across multiple channels, all brought together on our homepage, making it even easier to follow our latest conversations from your favourite platform, whether YouTube, Spotify or Apple Podcasts.

But this is not all…

4D Azores Gathering – Fall 2027

What happens when 4Ds gather in the middle of the Atlantic?

Join us for the first 4D Azores Gathering, taking place in the Azores in Fall 2027.

Many feel Azores is like Jurassic park. Possibly the best setting for the types of conversations we will have.

More details will be announced soon. In the meantime, if you have any questions, please feel free to contact Ida Helena Rust at: ida@humansignature.org.

Want a glimpse of Azores, check it here:

https://www.visitazores.com/en

First Book Club – Slow School House – September 2026

If you are looking for an inspiring book to end the summer with, or to read before getting caught up in the daily work routine again, consider Eric Ries’ Incorruptible.

More information soon!

We hope you find these developments exciting and looking forward your active participation.

See you on August 27th!

Can’t wait,

Heraldo
on behalf of the 4D Community

PS: Last Saturday I had the pleasure of publishing my conversation with Ida Helena Rust as part of Playing the Second Half.

If you would like to get to know Ida from a more personal perspective before our 4D Conversation on August 27th, I strongly recommend that you watch or listen to the episode below.

Playing the Second Half

Playing the Second Half - Conversation #11: The Adventure of the Spirit

Wow! Our Episode 8 takes us to the Azores for a fascinating conversation with Ida and The Adventure of the Spirit…

Listen now

3 months ago · Heraldo Sales Cavalcante

As many of you know I am an avid hiker, and I believe it transforms me as a human. The more I read about it, the more I have concluded that the modern way some people (like me) disappears into the wild, sleeping rough, freezing, and sometimes even starving on a long thru hike, without even having a clear purpose the more I wonder how similar I actually am to a medieval pilgrim.

I have done three extended hikes, 2022, 2024 and this winter as well. The first two were by foot, and this winter we did it on skis. The total length of my journey each time was around 1300 days, and the time we were out with my wife was between 53 and 60 days each time. It has taken us through the Swedish mountains from Grövelsjön to the northernmost tip of our Sweden.

Our latest (winter) pilgrimage through the Swedish mountains.

Let us say it is addictive, and I do not believe it has to do with either the beauty of nature or the physical feat of doing it. I think it boils down to how it changes you as a human. I think this is something we have been doing throughout times for all types of reasons. Let us call it pilgrimage.

I want to tie this to the phrase Solvitur ambulando, which is a Latin phrase attributed to Diogenes of Sinope and means “it is solved by walking”, which often taken as a way of getting a solution to a problem with practical means instead of just by thinking about it. Applied to Pilgrimage or modern day thru-hike it means that there has to be an underlying problem that needs to be addressed.

For a medieval person it may have been to get eternal salvation in religious terms, but research (and contemporary writing) seems to indicate a manifold of other reasons such as:

· Adventure – if you had the means a long journey was simply a way of seeing the world; if you had means it would be a journey to Jerusalem, but if you had less, it would simply be a few weeks of walking to the nearest cathedral.

· Economy – a pilgrimage could be a way of a way to get new revenue by selling goods you have made, entering into service, or procuring goods you needed for your business.

· Escape – many times people living in dire circumstances or under threat of being married or being ousted from your community was a reason to leave for a pilgrimage.

· Social – a pilgrimage was often a highly social experience, you often travelled together for safety anyway and you met at inns and taverns. Nowhere else is the clearer expressed than in “Canterbury tales” by Geoffrey Chaucer.

· Status – a pilgrimage was something that gave you status, which is expressed very well in all the little souvenirs that the pilgrims brought home.

Of course, every pilgrim had their own motive, which could be a combination of some if not all of the above. For many the religious motive was the most important, but they still had the added benefit of one or many of the ones mentioned above. As a matter of facts we can argue that the pilgrims to some extent where similar to the modern day tourists with the amulets being replaced by tacky souveniers.

So, what I am here to discuss is how a modern thru hiker find their motives above. First let me focus on the non-religious ones and then skip to what we see as “the secular religious motive”.

It is very easy to understand that all of the above may apply to a modern hiker. They seek adventure and many see the social part of it, it probably gives status, and at least I have met people escaping a home situation for instance after a divorce. Completing it gives you status (a badge, a mention, or at least cheers on social media).

I think it is rarer for having it to support your economy, but there are those who manage to monetize their stories either on social media, by writing a book or simply telling their stories on various social events.

But what replaces religion for a modern pilgrim. We are not seeking out shrines of holy bones, we simply walk from A to B. I believe that it goes back to Solvitur ambulando, not simply by walking but by disconnecting and being away from all day-to-day problems. Out on the trail you are away from all the updates, the 24-hour news cycle, the discussions with your colleagues and your normal job.

There is also a sense of equality on the road, hierarchies are mostly gone, and you great any fellow traveller as an equal.

I think the growing interest in long hikes and long journeys using your own body is directly connected to the rise in social media, not only because of all the updates from others but also the opportunity to experience the same experience as a medieval pilgrim.

As a matter of fact I know that a pilgrimage transforms you, and I am sure exactly the same applied to medieval pilgrimage to Canterbury, Santiago di Compostella, Trondheim or Jerusalem, and maybe the more we get connected and integrated with AI and social media, the more we need our own pilgrimages.

Let me end with a piece of poetry I wrote a while ago, maybe explaining my own experience better than I can express in prose.

Solvitur ambulando

Ahead is marked by cairns,
the path
meanders, crawls
beneath the bluffs, across
a frothing stream
over rocks while
step
by step
inside we meet
ourselves afresh.

Later, guided by
the breeze
our shadows follow
ridges diving into
valleys
never lost as we follow
a direction laid before.

We do not talk,
but listen
to our footfalls, heartbeat
breath and birds.
we watch the wind
and taste the scent of soil.

At break we gaze
at clouds
you point and say
“I see an elephant”
but I only see a pig
chasing a princess
but never tell.

At night
we listen to the sky,
our tent, the home we carry
hugs us gently
as we drift into a sleep.

We wake to rain
lashing, and in a hurry
we have to break
the camp and hurry onward
as we have miles
to go before we rest
and all is solved
by walking.

All words are my own, but I used Chat Gpt to do some research on pilgrimage ant it ties to Chaucer and modern day thru-hiking.

Source: Author as prompted in ChatGPT 5.5 Instant

By Alon Rozen, 4D contributor, Dean and Prof of Innovation, École des Ponts Business School

Allow me to start from the end and then circle back as a TL;DR: 1) We did this whole AI thing a** backwards! 2) Now it’s time to try again and get things right.

The Pitch

The coming AI backpedal will not be a retreat from AI. Neither people nor companies will suddenly decide that generative AI, agentic AI, coding assistants, copilots, knowledge agents and all the rest are useless toys/tools. That would be absurd. AI is useful. Very useful. Sometimes stunningly useful.

The Great AI backpedal I’m referring to will be a massive, pervasive and stunningly fast retreat from naïve AI deployment, not from AI itself. It will be a movement away from grandiose CEO declarations, tokenmaxxing, job replacement delusion, vibe-coding for fun and profit, credentialed claws, unbounded agents, open-ended AI budgets, zero accountability and, in general, the “we are all in on AI” woohoo enthusiasm.

Actually, it is not just one backpedal I think we are going to see, but three simultaneous strategic redeployments:

1. Economic backpedal: from “AI at any cost” to “AI has to at least pay for itself.” Crazy as it may sound, there will be budget requests and allocations, project attribution, a need to demonstrate ROI, and both accountability and affordability will need to be justified.

2. Managerial backpedal: from “You must put AI everywhere or else...” to “Let’s think carefully about where and how to deploy AI to create value.” No more implementation by decree and delusions of AI everywhere for everything. Instead, a move to questioning how to redesign processes to make effective and efficient use of AI. And, oh yeah, the realization that we probably need an AI strategy while we’re at it.

3. Technical backpedal: from “Let’s find the best LLM that will solve everything and do all the work” to “We should consider what AI models we should use for different jobs and how to implement it.” No more trying to guess the frontier model winner and paying a premium for ephemeral and theoretical performance gains. Rather moving to “fit-for-purpose” thinking, adopting vertical solutions when available (SLMs), and finding the least expensive models we can use that get the job done. This probably means moving from American closed models to Chinese open models for most work and using frontier models for very specific use cases that justify the extra cost.

Here are some fun AI facts that may explain some of the backpedaling I expect:

• By some accounts most tokens generated by LLMs are to correct content, bugs or friction created by the LLMs themselves. Ouch! (up to 80% according to @Aiswarya_Sankar on X, 26 May 2026)

• Weaker AI models fix their own mistakes better than stronger ones — 26.8% vs 16.7% intrinsic self-correction in one cross-model study — because the strongest models make fewer but “deeper” errors that resist fixing. So much for the comforting idea that we can simply have AI check AI. (Source: Y. Li, Decomposing LLM Self-Correction, 2026, arXiv:2601.00828.)

• About 80% of workers believe AI saves them as little as 0 to a maximum of 4 hours per week, while 80% of CXOs believe AI is saving them a minimum of 4 to more than 12 hours per week. Why this massive perception/performance gap? (source: L. Ellis, WSJ, 21 January 2026)

• Even when AI is wrong, people follow it ~80% of the time, performing “worse than having no AI at all.” People (which include the employees and managers mentioned above) not even checking AI outputs is a form of dangerous “cognitive surrender”. This can’t end well. (Source: Shaw, S. D., & Nave, G. (2026). Thinking—fast, slow, and artificial: How AI is reshaping human reasoning and the rise of cognitive surrender [Preprint]. PsyArXiv. https://doi.org/10.31234/osf.io/yk25n_v1)

• Mentions of AI’s business impact on S&P 500 earnings calls have risen 3–4x since 2023, but most companies still haven’t reported measurable results. A case of reality not living up to the hype? (Source: Azeem Azhar Exponential View research, 25 June 2026)

Let’s be clear, this is not an anti-AI article. Quite the opposite. I use it. My team uses it. My students use it. My children use it. In some areas it is already a productivity booster, a creativity amplifier, a research accelerator, a learning companion and, when used well, something close to a cognitive exoskeleton that allows me to learn and work faster than ever. It allows me to explore my curiosity in ways I never could. In my own work, it has become part of my “thinking stack” (the ThinkingOS for those of you who read that post), and increasingly my intellectual sparring partner for posts like this. And, of course, it has helped me refine my thinking about this article too!

My point is that I think we are now seeing AI for what it is beyond the usual hype cycle of new technologies. After peak exuberance, following Gartner’s infamous model (the Gartner Hype Cycle), we will now quickly move toward something much less cinematic but much more pragmatic: efficiency, governance, accountability, affordability, workflow redesign, unit economics and much more thoughtful deployments. In other words, the enterprise AI reversal will not be from adoption to rejection. It will be from magical thinking to managerial accounting. The euphoria of easy wins is fading, now we realize you need to put in the hard work to reap the benefits. Just like anything in life.

And yes, I realize managerial accounting does not sound particularly exciting. But it is where innovations go to become real and impactful. For those who haven’t done an MBA (recently), managerial accounting is the process of measuring, analyzing, and communicating financial and operational information to help managers allocate resources, evaluate performance, and make better strategic and operational decisions. In other words, the same best practices we have applied across the board – project management, clearly-defined budgets, timelines, milestones, KPIs and accountability... they are the next step. Nothing radical, just applying what we already know and do for everything else to AI.

The end of the “all in” phase

Reality check: for the last three years, no CEO (or founder or investor) really wanted to sound cautious about AI. The battle cry “We are all in on AI” became the safest sentence in the corporate world. No need for a strategy, just a declaration and a profession of faith. It was bold, vague, reassuring and fashionable all at once. It did not matter whether the company was in education, insurance, banking, consulting, construction, luxury goods, software, logistics, or frozen pizzas. Everyone was, by declaration, “AI-first,” “AI-powered,” “AI-enabled,” “AI-native,” “AI-augmented,” or at the very least “actively exploring AI.” Anyone who lived through the dotcom era recognized the zeitgeist. Not sure anyone knew what those terms actually implied but, darn, it sounded good and boosted stock prices and valuations.

Many CEOs demonstrated their conviction by issuing “use AI or else!” statements, forcing AI down the throats of their employees. Some CEOs I know even proudly told me that they were firing any staff that did not adopt or refused to adopt AI.

To be fair, there was a certain logic to this. AI looks (or looked) like a once-in-a-lifetime technological shift. No CEO wants to look like the person who looked at the internet in 1996 and said, “Interesting, but probably not relevant to us.” Or the smartphone in 2010 for that matter. There is a reason that corporate language shifts quickly around major technological transitions. The language is partly descriptive, but it is also performative.

But there is a huge difference between performative awareness and magical thinking. “All in on AI” is not strategy or even a strategy. It is a posture. It is useful for signaling urgency, but not very useful for deciding what to build, how to build it and why even to build it. Just like innovation, AI needs a strategy. Rudderless innovation, rudderless AI. And that is probably what the first phase of AI deployment was most naïve about. Defining a strategy is hard work and takes time. It requires objectives, tradeoffs, real problems to solve and creating value that the market needs, wants and is willing to pay for.

AI’s Uber moment

Besides the hype and marketing fueled by private equity flowing in unparalleled quantities to a small cadre of huge AI players, one of the reasons this “AI problem” has taken time to surface is that we have not yet had a clean view of the true cost of AI. This is where Uber is useful as an analogy.

For years, Uber did not teach us the true cost of urban transportation. It taught us the price of transportation when investors were willing to subsidize the ride. Consumers learned to love the price before they understood the economics. Cities learned to adapt to the service before they understood the externalities. Investors learned to celebrate scale before profits mattered quite as much as they do today.

AI may be having a similar moment.

For the last two years, many enterprises have been trained to think that machine intelligence is cheap, abundant and almost frictionless. A monthly subscription here, an API there, a few agents, MCPs and voilà — the organization begins to think that AI is simply another layer of software. But the current price of AI is not the true cost of AI. It is the price under conditions of significant subsidy, strategic land-grabbing, investor optimism and market capture.

The real question is not whether AI is cheap. The better question is who is currently footing the difference between what we are paying to use it and the actual cost of providing it? And what happens when we start to pay the real price?

The mass realization that the difference between the actual price of AI and the cost of AI has been absorbed by venture capital, hyperscaler partnerships, model providers trying to acquire users, and companies racing to become the default intelligence layer for work cannot last forever. Heck, it looks like it won’t even last much beyond the IPOs of OpenAI and Anthropic. The writing is on the wall - infrastructure costs, investor pressure, cloud bills, energy constraints and simple financial reality will force AI providers to charge customers something closer to the true cost of the service. And when that happens, some enterprises may experience significant and painful “AI sticker shock”.

Even a hyperscaler like Microsoft is part of the backpedal. From a “one model to rule them all” naïve deployment with just OpenAI to run Co-pilot, Microsoft is moving to a multi-model approach spanning its own in-house models, Anthropic, and — per recent reporting — Chinese open models like DeepSeek and Kimi for selected Copilot tasks. From betting on premium pricing and performance to fit-for-purpose and budget. A sign of things to come.

And here Uber is again an interesting case study, but not for the same reasons as above. Multiple reports announced (including TechCrunch) that Uber was surprised to discover that it had blown its full-year AI budget for 2026 in only 4 months. And, Microsoft, to stay with the same examples, reportedly curtailed access to Claude Code for engineers because of cost pressures. Reuters Breakingviews (R. Mak, 2 June 2026) recently argued that corporate AI “sticker shock” will force restraint, noting that pay-as-you-go AI bills are harder to track than traditional software subscriptions. These are early signals of this shift.

One or two anecdotes does not a trend make. But several anecdotes pointing in the same direction are at least a weak signal. And weak signals are often where the interesting work begins.

The first phase of enterprise AI, the tokenmaxxing phase, was priced for accelerating adoption, grabbing market share, and getting people and companies to use AI. The next phase will be priced for performance, meaning costs and benefits need to at least cancel out.

A fair objection deserves a hearing here: the true cost of AI is probably not fixed and may be even falling quickly. Inference cost per token has been dropping by roughly an order of magnitude a year, open-weight models keep lowering the price floor, and prompt caching helps cut effective costs even further. If the real cost is collapsing toward today’s subsidized price, perhaps the sticker shock is transient, and the enthusiasts turn out to be retroactively right. I take the point — but it actually reinforces the point I’m trying to make. A fast-falling cost curve is precisely why disciplined, fit-for-purpose deployment wins: the organizations that meter usage, match models to tasks and avoid locking themselves into premium frontier model pricing are the ones best positioned to ride the curve down. Undisciplined “all in” spending doesn’t beat the curve, it front-runs it at the worst possible price. Over-paying for under-performance is not something a CEO wants to be asked about in an earnings call.

The SaaS reality distortion field

A large part of the problem is that companies are still trying to understand AI through the mental model of enterprise software. Traditional SaaS trained managers to think in relatively predictable categories: users, subscriptions, annual contracts, and renewal negotiations. Once you bought the tool, the cost was reasonably visible. Not always low, of course, but at least legible. AI broke that model.

The magical-thinking formula is simple:

AI cost = subscription price x number of users

The reality turned out to be something much less clear or comforting:

AI cost = users x workflows x model quality x token volume x context window x latency x verification x integration x governance x error handling

In other words, instead of paying for limited-cost unlimited-use software with guaranteed performance that you can easily attribute to projects and users, you are paying for limited usability at unlimited-cost that you can’t easily quantify, qualify, attribute to and for which you have absolutely no performance guarantees. That sounds like a potential train wreck in the making and something that no reasonable Board will allow to go on.

This is why the cost surprise can be so painful. A SaaS tool usually becomes predictable once purchased. AI can become less and less predictable the more successful it becomes. If employees use it more, costs rise. If workflows become more complex, costs rise. If context windows expand, costs rise. If agents perform multiple steps, costs rise. If the organization moves from casual prompting to production deployment, costs rise again.

The most dangerous illusion in enterprise AI is mistaking the access cost for operating cost. It is the same kind of fantasy that makes people confuse the sticker price of a car with the full cost of owning, maintaining, insuring, parking and fueling/charging it. The purchase price is visible. The total cost of ownership is not.

The new doctrine: informed AI deployment

None of this is an argument against AI. It is an argument against lazy AI deployment and for a thoughtful AI strategy.

The next phase of enterprise AI will belong to companies that become disciplined fastest. Not the companies with the most breathless declarations. Not the companies with the most pilots. Not the companies with the most internal demos. The winners will be the organizations that learn how to deploy AI with informed restraint, meaning they will use it generously for the use cases that generate a positive ROI and sparingly for the other use cases.

First, companies need an AI strategy! They need to think carefully about why, how and what to build with AI. As well as thinking about how to avoid cognitive surrender in key aspects of the business, especially mission critical aspects (see some of Benoit Bergeret’s writing/thinking on this, or my previous article on the ThinkingOS).

Second, AI needs to be reconsidered as infrastructure, not software, and budgeted accordingly. AI should be properly treated as a variable production input, not erroneously considered as a fixed subscription tool.

Third, companies need to think about their AI KPIs. In a nutshell, how to measure the cost per a reliable outcome. Not cost per user, not number of tokens, not number of agents. This also probably implies companies inventing their own AI-relevant metrics that measure cost per some form of usable, trusted, repeatable, business-relevant output. For example, a support team’s real metric isn’t cost per query but cost per resolved ticket — one that stays closed, with no human escalation. An engineering team’s isn’t tokens burned or “percentage of code written by AI”, but the fully-loaded cost per shipped feature. A knowledge worker could measure cost per shippable draft that requires little or no editing.

Fourth, move to multi-model thinking. Reserve frontier AI for frontier-value tasks. Not every task deserves the most expensive model, and people need to learn AI discernment. Smaller and/or open models, sharing skills, redesigning workflows or just relying on human judgment may be both better and cheaper.

This is where the China story stops being a footnote and becomes integral to the backpedal. The steepest part of the falling AI cost curve is being set by Chinese open-weight models — DeepSeek, Kimi, Qwen — which now rival OpenAI and Anthropic on many everyday tasks at a fraction of the token price. This is no longer fringe: even Microsoft in its multi-model approach is reportedly weighing a move of selected Copilot workloads to Kimi and DeepSeek, with estimated cost savings cited in the region of 90% (!) for routine work. For most companies the question is quickly shifting from “which frontier model?” to “why are we paying frontier prices at all?” The fit-for-purpose default will be increasingly open and Chinese. There is a catch, though, and it is the same one that haunts the rest of this post: those Chinese prices are themselves strategically subsidized, based on distilling the frontier models, and arguably subsidized even more aggressively than the US models. The true cost of AI is ever harder to discern.

Fifth, start designing human-AI systems, and stop thinking about how to replace people with AI. As I have mentioned elsewhere, and hat tip to Antoine Fourmont Fingel, negative productivity (reducing the number of people you need to run your business) is less interesting and achievable than positive productivity (doing new, more and better with the people you have). The real unit of transformation is not the model, or choosing the right model. It is the redesigned workflow. Now that the delusion of replacing everyone with agents is dissipating, the hard work of rethinking human-AI interactions to unlock new workflows, new business models, and new ways to create value can begin in earnest.

Technology, yes. Magical thinking, no.

I have spent much of my professional life arguing that leaders need to understand technology better. Not necessarily to code. Not necessarily to become engineers. But to understand enough of the “hardish” logic of technology to make better strategic, organizational and human decisions. That remains true. In fact, AI is making it truer every year.

AI cannot / should not be delegated to the “techies”, a consulting firm or a handful of Gen Z employees who are good at prompting. All executives and managers, right up to the board need to become at least AI-aware, if not fully AI-literate.

But tech-aware does not mean tech-worshipping. We need a more mature and a more balanced approach. Keep the best practices of business that have passed the test of time, while adding, responsibly, the new opportunities that AI offers. The humility of understanding that when we introduce AI we don’t remove the complexity of our people, our businesses and our ecosystems, we are changing relationships, incentives, workflows, responsibilities, cost structures and sometimes identities. Change management is extremely difficult; it doesn’t get easier with AI. On the contrary.

This is why the next phase will be so interesting. The companies that do best will not be the ones that shout “AI-first” the loudest. They will be the ones that ask better questions earlier: where does AI really create value, where does it create activity masquerading as value, where does it lower cost, where does it shift cost, where does it increase speed, where does it increase fragility, where does it augment judgment, where does it outsource thinking, where does it help people become better, and where does it make them lazier?

No easy answers to these questions. But that is precisely the point.

I’m convinced that by this time next year, AI will become deeply integrated into the enterprise but in a way that is much more governed, metered, selective, embedded and accountable. It will be budgeted like infrastructure, managed like risk, evaluated like CAPEX and deployed like organizational redesign.

The Great AI Backpedal is not the end of enterprise AI. It is the end of pretending that AI is magic.

But that is not a sign of failure. It is what happens when a technology grows up. The AI revolution is not ending. It is simply entering the age where hype fades and unit economics ensure a more reasonable re-deployment.

The king is dead? Long live the king!

Thanks for reading.

AI usage:

• Claude Opus 4.8 High was used to critique the article and suggest edits.

• ChatGPT 5.6 Sol was also used to critique the article.

• ChatGPT 5.5 Instant was used to generate the cover image.

Anna’s main focus of concern is disinformation and the fact that we are losing our sense of truth.

We discussed many topics, including raising two children in this world and her concerns about helping them become truly good citizens.

Enjoy, and let’s continue the discussion on understanding the metacrisis, not only from an intellectual and rational perspective, but also as individuals.

Alon shares why the metacrisis is less about a dictionary definition and more about a distinct feeling of “ick” amid the realization that we are facing a barrage of crises with no values-based leadership to guide us forward.

Together, they also dive into:

· The “hamster wheel” of AI: Why the rapid pace of technology is fueling a backlash against techno-optimism.

· AI as a Socratic tutor: How to avoid using AI as an “intellectual crutch” and instead use it to ask deep questions that sharpen your own thinking.

· The complex rise of the “Digital Twin”: When you build an AI agent that thinks, writes, and acts exactly like you on the job, who owns it? Alon explores the complicated legal and ethical future of our virtual selves in the workplace.

Please also read Alon’s 4D Community excellent posts on AI and doubt, as well as watching his in depth 4D Conversation on complexity.

Your Mind is Not a Prompt: How AI is hacking your cognition stack… and how to hack back!

Doubt is a Superpower. Here’s How to Use It!

4D Conversation: Decoding Complexity: From Concept to Reality

Here is a transcript of the conversation:

Nathan Hegedus

Welcome to 4D Conversations. I am Nathan Hegedus, and I am here with Alon Rozen, the Dean of the École des Ponts Business School in Paris.

Alon is a fellow member of the 4D community. He has written two powerful pieces on AI and cognition and the power of doubt, and he also hosted a 4D conversation on complexity. Today, we are continuing our series exploring the question, “What is the metacrisis?”

So, Alon, welcome. What does the metacrisis mean to you?

Defining the metacrisis as a captainless ship

Alon Rozen

Hey Nathan, thanks so much for having me. I am really excited to have this conversation, especially because the 4D community is pushing this dialogue forward. Hopefully, we are creating ripple effects by hosting these discussions and helping others understand why the metacrisis is such an important subject to tackle.

If you had asked me what the metacrisis was even six months ago, I probably would have given you a standard dictionary definition—multiple, overlapping crises happening at the same time, where one finishes and another has already begun.

But the true feeling of “ick” that defines the metacrisis for me is the sense that while we face all of these crises, no one is driving the ship. There is no captain, and the leaders we do have are so devoid of values that there is no way to trust them to right the course.

It is a feeling of dread, like being hit by an asteroid attack where multiple rocks strike the Earth simultaneously, and there is no trusted figure to help us find a way forward. Nobody speaks with both true confidence and a genuine constituency.

We see these overlapping crises everywhere—the war in Ukraine, conflicts in Israel and Iran, shifting economic situations, the ongoing cold war between China and the US, and the questioning of historic alliances like the one between Europe and the US. Everything feels up in the air, yet we do not see world leaders who merit our respect, speak truthfully, or offer a path forward. That is why the metacrisis feels so heavy. It is hard to remain optimistic when we are attacked on all fronts without a values-based backboard to lean on.

Nathan Hegedus

That makes total sense. I often call it a sense of overwhelm because it is so difficult for people to get a handle on it. What do you think led us here? How did we end up in this situation?

Technology and the hamster wheel of permanent overwhelm

Alon Rozen

I think we can see a clear parallel in how people view AI today. There is the “wow” of technology alongside the “shit” of technology—a feeling that we simply do not have a handle on it, that it is bigger than we can manage.

The current AI backlash is very much a backlash against techno-optimism. It is a reaction to the feeling that progress is moving far too fast to keep up with. Whether you are a politician, a business leader, an employee, or working in a government role, everyone feels like they are in a state of permanent overwhelm. You feel like you have to study AI twenty-four-seven just to keep up with a model that came out two days ago, and you will still never catch up. It is a hamster wheel.

When you combine increasing complexity and opacity in our economies—from the decisions leaders make to the black-box technologies we use—we lose the script. We already lost it with smartphones because we do not truly understand how they work, but with AI, we have no clue what is piloting our very thoughts.

This is another layer of the metacrisis. It is an existential uncertainty where technology moves too fast, and instead of controlling it, it feels like it is controlling us. Everything is interconnected, which is the definition of complexity—a system of interconnected parts that produces its own emergent traits. We are living that reality right now.

Nathan Hegedus

During your previous conversation on complexity, there was a lot of discussion about how, to some degree, systems simply have to crash. How does that square with rational optimism? Some listeners might argue that climate change is crossing irreversible lines and inequality is too severe to ignore. How do you reconcile those two views?

The ozone hole precedent and the case for rational optimism

Alon Rozen

It is a difficult question with no easy answer. While complexity denial and climate denial still exist, it is becoming increasingly difficult to stay in those camps. Even the people downplaying it probably do not sound as convinced in private as they do in public interviews.

For me, a great metaphor for rational optimism is the hole in the ozone layer. That was the last major global challenge where humanity truly came together. Anyone old enough to remember the seventies, eighties, and nineties remembers how we talked about it constantly, warning of skin cancer and environmental collapse.

But humanity got its act together and closed the ozone hole. It is a remarkable achievement. Perhaps we need an issue as highly visible as the ozone hole was to force us to unite. Once climate change becomes completely undeniable, the consensus and desire to act can have a rapid, massive impact. We phased out chlorofluorocarbons and saw the ozone layer begin to heal in less than a generation.

That gives me hope that we can act before our systems reach a point of no return. I still believe we can.

Nathan Hegedus

That is a fantastic example. Having lived through that era as well, I remember people talking about the ozone layer with the same sense of helplessness and irreversibility that we feel around climate change today.

That brings us back to rational optimism and taking a beat to stay calm. You recently wrote a brilliant piece about AI. How can people approach AI in a way that actually works for them?

Moving beyond intellectual crutches to use AI as a Socratic tutor

Alon Rozen

If you do not think intentionally about how you use AI, you can actually become dumber without realizing it. If you constantly rely on AI for shortcuts because you are short on time—simply accepting whatever text it generates because it looks good enough—you end up walking on intellectual crutches. You risk reaching a point where you can no longer write or think deeply without it.

On the other hand, if you use AI correctly, it is the most extraordinary tutor you could ever ask for. It can even teach you about yourself.

I suggest people try a Socratic approach. When I want to write an article, I will tell Claude or ChatGPT, “I am planning to write a piece on this topic, and I want you to ask me a series of questions to help me think it through. What am I missing? Is there a better angle?”

I have had these Socratic conversations go on for hours. You would be amazed at the quality and sharpness of the ideas that emerge. If the AI tries to sneak in answers, I tell it, “No, do not answer for me. Just ask the next question.” It forces you to play an active role, helping you articulate your own genuine thoughts rather than outsourcing your brain.

It is just as powerful as a great real-life conversation where you bounce ideas back and forth. You can even do this with your own drafts. Instead of asking the AI to rewrite your text, ask it, “Where are my arguments incoherent? What gaps am I missing? Am I writing for my audience, or just for myself?” It is a wonderful, fascinating way to sharpen your thinking.

Nathan Hegedus

I have done that intuitively, but explicitly prompting the AI to only ask questions is an incredibly inspiring way to structure it. Is there anything else you want to touch on before we wrap up?

Redesigning workflows and the complex future of digital twins

Alon Rozen

AI is the absolute elephant in the room. On an individual level, it is forcing us to redesign our own creative and analytical processes. We are actively deciding how we want to integrate technology while keeping our unique voice intact. We have never had to do that before.

On an organizational level, we are moving past a naive deployment of AI where companies just slap it on as a band-aid to do things faster and cheaper. Instead, thoughtful companies are starting to redesign their processes entirely. They are asking, “How can we use AI to improve customer satisfaction?” rather than just “Where can we use a chatbot?”

For example, instead of a robotic automated response, an AI could tell a customer, “We received your message. Our team starts at nine tomorrow, your email is third in line, and Jane usually responds within ninety minutes.” It feels incredibly personalized, yet it requires very little operational change.

The third area, which is more complex, is personal identity. Let us say you become an AI power user and build a digital twin at your company. You train this agent on everything you have ever written so it thinks, writes, and speaks exactly like you.

One day, the company might realize they have a digital twin of Nathan. They might decide they do not need to pay you anymore because they can just keep your digital twin.

We might see a future where employees actively hide their personal AI usage, keeping their custom models outside company borders so their employers cannot claim ownership over their intellectual identity. Conversely, companies will have every incentive to capture that data. This is going to make for some fascinating legal and ethical cases. It sounds like science fiction, but reality has a habit of outrunning fiction. We must figure out how to separate the individual, the knowledge, and the agent.

Nathan Hegedus

I can easily picture a future where people have different versions of their digital twins floating around at every company they have worked for.

It also feels like the thoughtful, calm, and optimistic approach you take toward AI is a microcosm of how we might navigate the broader metacrisis.

Directing AI toward global solutions and positive use cases

Alon Rozen

I agree. On a macro level, we can find incredibly positive use cases for AI, like using it for drug discovery or synthetic research. At our school, we are working with an Oxford professor on running experiments using synthetic data to identify which trials are actually worth pursuing in the real world.

If we align individual development goals with strategic development goals, we can use this computational power to solve major systemic issues. For instance, how do we feed the world without carbon-based, oil-derived fertilizers? That is a massive calculation problem that AI is uniquely suited to solve.

That is where my rational optimism comes back. We can focus AI on positive global impacts and regulate its more nefarious uses, like military drone targeting. We are at a red pill, blue pill moment. Do we use this technology for good, or do we allow it to be used destructively? We are currently split, but I hope we choose the path of positive potential.

Nathan Hegedus

That is a beautiful note to end on. Thank you so much, Alon, for this wonderful conversation. I look forward to reading your upcoming pieces. We will include all the links to your current work in the show notes.

Thank you to everyone for listening. We will be back soon with another 4D Conversation. Take care.

Alon Rozen

Thanks so much, Nathan. Thanks to everyone for listening. Bye-bye.

In this episode I had the privilege of talking with legendary management thinker Henry Mintzberg about his ideas on rebalancing society. We do not come to any hard conclusions but, as someone who did grassroots work on building civil society once upon a time, I love his insight.

And after we talked, I was deeply inspired by a presentation from New_Public, in which they see opportunities (at slide 40) for AI to actually make it easier to organize and coordinate sustainably without everyone burning out. I’ll have to follow up with Henry on this!

Here is a transcript of our conversation:

The Balanced Society

A 4D Conversation with Henry Mintzberg

Nathan Hegedus Welcome to this edition of 4D Conversations. Today I am with Henry Mintzberg, one of the most influential management thinkers alive. His work on strategy, organizations, and leadership has shaped how we understand how institutions work in reality. Later in his career, he turned his attention to society itself and wrote Rebalancing Society: Radical Renewal Beyond Left, Right, and Center.

Henry, welcome and thank you for joining.

Henry Mintzberg It is my pleasure to be here, Nathan. Thank you.

Nathan Hegedus I thought I would start with the through line for your career, because your career started with a very hard look at how organizations actually work. What is the connection between that work and the rebalancing society work that you are doing now?

Henry Mintzberg The first thing I looked at, even before organizations, was how managers actually work. Sometimes people call me a contrarian, but I am not a contrarian at all. I don’t take counter positions for their own sake; I take them because the popular position just seems wrong to me. If I read textbook definitions saying that managers plan, organize, coordinate, and control, but then I sit in their offices and nothing they do looks like any of that, I say so.

In 1991, I visited Prague, two years after the fall of the Berlin Wall and the Velvet Revolution in Czechoslovakia. This popular idea that capitalism had triumphed just seemed dead wrong to me. Capitalism didn’t triumph. Balance triumphed.

It wasn’t just about balance across two sectors, either. I wrote an article a year later making the point that it was the importance of a community sector that established true balance. Ever since then, I’ve looked at American history more closely. The secret to American success over the last 200 years was not just business being effective and government being constrained. It was also community activities that were absolutely key to the development of the country. They were just as important as business activities, something Alexis de Tocqueville wrote about back in the 1830s.

Eventually, I came to call this the plural sector. You don’t just have public and private, or economic and political—you have the social sector. I call it plural for two reasons. First, there is a huge variety of organizations within it. They are either owned by their members, like co-ops, or they are owned by nobody. When people ask what it means for an organization to be owned by nobody, look at Greenpeace, McGill University, or Harvard University. They are trusts of one kind or another. It is a massive sector, and using the word “plural” helps people see it taking its rightful place alongside the public and private sectors.

Nathan Hegedus Does the current moment seem different to you? There is democratic backsliding in places, the rise of AI, and we talk a lot at 4D Conversations about the meta crisis. Does the reality of 2026 change anything in the way you see things?

Henry Mintzberg You could say I saw it coming—not in these exact details, but one thing was quite clear. Capitalism became the be-all and end-all of everything in the West, and everywhere else except for a few communist countries. It was only a matter of time before that went out of balance, just the way communism went out of balance.

Under communism, business was weak, and communists did a great job of suppressing absolutely every plural sector community movement they could because it posed a threat to their power. Today, I am amazed that people in the United States don’t get it. They just don’t see that the power of business has become absolutely, unbelievably lopsided.

Nathan Hegedus And why do you think that is? What happened?

Henry Mintzberg It has been a long time coming. If you look at history, business has always sought to be dominant in the United States. There were battles right from the beginning—there is even a letter from Thomas Jefferson warning about business trying to take over the country. That advance moved steadily.

However, the plural sector used to be very strong in the United States. As de Tocqueville described, America was a deeply community-driven place. That community element weakened for a number of reasons, but I think the main culprit is technology. Every new technology that has come along has driven us out of community and deeper into ourselves.

When the automobile came along, you didn’t walk on the street or see your neighbors anymore. It’s like putting police officers in a squad car versus having them walk a beat—one is inherently more communal than the other. Then television arrived, and we ended up watching things at home alone. People stopped interacting socially.

I use a joke about modern highways. When somebody tailgates you on the highway, their attitude is “get off my road.” It is a deeply individualistic environment. But who tailgates you on a sidewalk when you are walking? Someone who is reading an iPhone. They bump right into you because they aren’t even looking at you. It is technology all the way.

Nathan Hegedus I am interested in the fact that the plural sector seems to have trouble organizing itself. Why do you think that is? In my own writing, I’ve looked at the work of Theda Skocpol and the fact that the modern plural sector in North America is run almost entirely by professional managers and professionals. Are they just unable to agree, or is the high level of professionalization an issue on its own?

Henry Mintzberg They certainly have a degree of competition with one another regarding fundraising and resources, which causes them to be wary of each other. But I think the fundamental issue is a question of mission.

The mission of Greenpeace is climate change. The mission of Doctors Without Borders is healthcare. The mission of Amnesty International is social justice. What they don’t realize, or don’t care to realize, is that they will get absolutely nowhere with any of those individual causes until we rebalance our societies. Because our societies are so lopsided, it fundamentally drives these problems.

Look at the COP climate conferences. They are “cop-out” conferences. They are jokes. Everybody flies around, comes home, and makes plans. But they don’t make four-year plans to match their four-year governments; they make 40-year plans. Then the next government comes along and cancels it. It would be laughable if it weren’t so tragic.

Nathan Hegedus So what is the work you are doing now? How do you address this?

Henry Mintzberg All I can do is write about this in a way that asks people to wake up and see what is actually happening. I am writing a pamphlet right now called Balance Now for the Sake of Survival, and I mean it literally. We will not survive this—whether it is wealth inequality putting us against each other, wars, climate change, or anything else. We will not survive a completely lopsided world.

Nathan Hegedus What would you recommend for the average person? How do they contribute to rebalancing society?

Henry Mintzberg I am mapping that out right now. People always ask, “What can I do?” But the “I” question must have a “we” answer. What I can do individually is get together with other people and work directly within our local communities to start changing things. That is the core of what I am writing about.

Nathan Hegedus Right, and I guess that is the trick. People get involved locally, but how do you actually scale that momentum up?

Henry Mintzberg I think the closest historical model we have is the Reformation. The framework I am using follows distinct stages:

1. Individual Engagement: You read, you research, you realize what is going on, and you wake up.

2. Communal Action: You get together with others to answer the “we” question. You engage in localized social initiatives.

3. Institutional Reform: As you advance, you begin to reform your institutions. You don’t start with central leadership or top-down institutional reform; you reform the institutions once you have built a massive underlying movement.

4. Global Consolidation: Finally, you consolidate those pieces globally.

Nathan Hegedus Do you have examples of countries or societies today that are operating with more balance?

Henry Mintzberg The Scandinavian countries and Canada are good examples, as are a lot of smaller nations functioning well—places like Costa Rica, Uruguay, and Botswana. They all face populist movements, but for the most part, they maintain balance. Canada, in particular, remains a very balanced place.

Nathan Hegedus Thinking back to de Tocqueville, he saw this immense, unique energy in American society. Is there any practical way to recover that energy today?

Henry Mintzberg I have a friend who is a business consultant in the States. He, another friend in Canada, and I get on regular calls together. He often tells me, “Henry, I feel so hopeless all the time, but when I get off these calls, I feel like there is hope.” Everyone needs to be able to find that feeling.

For a concrete blueprint of what this looks like, people can look up an article on my website—mintzberg.org/articles—about Coop Sapporo, a cooperative in Hokkaido, Japan. It is a phenomenal model for what I call a consolidated social economy.

Hokkaido, the big northern island, faces severe depopulation issues. When traditional businesses moved out because it was no longer profitable, Coop Sapporo moved in to fill the void. If a village doesn’t have a grocery store anymore, the co-op sends out mobile vans so people can shop. If there is no bank left in the village, they send out a van equipped with an ATM. They provide care from cradle to grave—handling everything from baby supplies and meals for school kids to funeral arrangements. They stepped in communally where the private sector failed.

Nathan Hegedus Wow, that is amazing. One last question, Henry: what, if anything, makes you genuinely optimistic right now about the future?

Henry Mintzberg Models like Coop Sapporo. My friend Jeremiah who finds hope after we have our conversations. And young kids who want to do something meaningful. There is always that energy out there. We just have to tap into it.

Nathan Hegedus Yeah. Well, that is great. Thank you so much, Henry.

As I continue to explore the idea of AuthorshIp (ownership) in AI, I can’t help but think of a 1917 series of photos of young ladies with fairies. These are now known as the Cottingley Fairies Hoax. What makes these “faked” photographs so disputed was that Sir Arthur Conan Doyle (yes, of the famed Sherlock Holmes) declared them real.

While this wasn’t the earliest photograph to be “faked” they are some of the best known. Photographic hoaxes have been around as long as photography has existed. So, what does this have to do with AI?

The question arose recently on the ethics of using AI on photographs? And whether or not the use of AI to alter a photograph constituted a distortion of “reality.” But then again - is Victorian Death photography a distortion of “reality,” after all the main subject of the photograph is no longer alive? [For a really interesting internet web rabbit hole - look up Debunking Victorian Death Photography.]

However, I digress, so back to the concept of “authorship” of photography.

The person who “presses the shutter and captures the photograph” owns the photo according to copyright law. Obviously, any contract dictates the full ownership if you are contracting someone to photograph you or something for you. This “new” law went into effect in 1978. This is why you do not have an expectation of privacy in a public place when it comes to having your photo taken.

Now with AI

US copyright law - which is kind of the defacto standard worldwide - has shifted to include definitions of authorship for AI altered images.

Photos have always been able to be manipulated. Somewhere in my vast portfolio of photographs of myself is one that a friend took in his office. There are two chairs in front of a window with a table in between. In the left chair I have on a jacket and am looking at the camera. In the right, no jacket and I’m looking at the left chair. This was done by manipulating the exposure.

When developing photos, photographers could manipulate the chemicals to get different a different look or feel from the printed image. Maison Francis Kurkdjian partnered with Christelle Boulé to photograph their perfume. The result is an amazing piece of art that forms when the perfume is exposed to photographic paper.

Now, what I find interesting is that prior to AI we had tools such as Photo Editors (I won’t name brand names) that allowed individuals to retouch photographs, or edit them with lenses, add layers with objects, etc. So the ability to recreate the Cottingley Fairy Hoax in the era of photo editors would be so much easier than in 1917 when the fairies were hand drawn and paper cut outs arranged to appear in the greenery.

Which brings me back to the question of AI or tool? Photo editing software has been around for years, and with AI it has gotten even better. The ability to use this software requires a degree of skill, much like painting a portrait requires a degree of skill, and some level of artistic ability.

Snapchat filters let you manipulate your own photo a myriad different ways. The skill is in the filter creator, not me snapping my picture - yet I still own the photograph - no matter which filter (or 13) I use. Now, Snapchat has kindly watermarked the filtered photographs, but with a photo editor - I could remove it.

So, who then owns an image or photograph that has been altered? All of the photos below have been mostly manipulated with either a physical lens held over the mobile phone camera, a setting on the mobile phone camera, or the photo software on the phone. None of this was done with AI - simply the tools available. So the copyright remains with me, as the photographer. But what happens when I’m not the photographer?

The Sale of a House Case Study

This is the same house. One picture has been “enhanced” by AI, while one has not.

While these pictures were taken 4 months apart, so the trees and grass had time to brighten up from winter to spring - the removal of the 20+ year-old tree (by AI) from the front yard was actually a direct violation of real estate code - as it presented a false picture to the buyers. Sure it gives a more direct view of the front of the house, but it is also a lie. This is a case where the manipulation of a photo should not have been done, as it falls under deceptive practices. Or if it was, then the listing also needed to also present a picture with the tree, with the image above noting it had been manipulated by AI.

I bring this up as this is a use case where the rules are very clear on what can and cannot be done. While the real estate agent could use AI to add furniture to a room, or present a different color scheme, the house had to be presented as it existed. Exhibit B has had a sky filter applied to it - which while somewhat deceptive, is not a violation of rules.

But what if I just wanted to take the photo (of my house) and add lightening or make it fall?

No rules broken here either. But note the subtle Gemini watermark. Sure, I could use a photo editor to do the same manipulations without the watermark - but it returns the the question of AI as a tool or AI as the Author/Artist.

Now, I am not the individual who actually took either of these photos in the beginning - so I have no ownership to the photos to start - even if it is of my house. So have I already broken copyright by using them without permission and manipulating them without permission?

The fine print in my sellers contract says I can use them - so the answer is no…. but had I picked a picture of a random house and done the same thing - the answer would be yes - maybe. This is where copyright becomes a huge grey area and authorship/artistry becomes a question of derivatives.

Because I started with one picture and altered it to be something different, one could claim that this is a derivative work. AI doesn’t know who owns the copyright/rights to an image that is uploaded, so it can’t help keep copyright from being broken.

At the same time, AI has been programmed to not create works of specific artists. It will create in the style of specific artists. I can prompt the AI to recreate the picture in the style of an impressionist painter with a blue starry sky in the background. And we’ll come pretty close to a Van Gogh painting.

For the one below I started with giving Gemini the image and asking Gemini to manipulate it in the style of Matisse and Kandinsky.

For fun I asked Gemini to create an image replicating the technique that Christelle Boulé used in her exhibition Expanded Drops (Third Image top row). Then I asked it to restyle it in the manner of Warhol, Kandinsky, Matisse and Delaunay.

All of this work to bring us back to the question of authorship. Who owns all these images? Well, technically - I do. Because I either took the photographs, did the editing or manipulated them via editor or AI. Although it’s not fully clear that AI created the images.

For this one I started by asking Gemini to create a photorealistic image of a yellow teapot with coral roses and hydrangeas. It took two tries to get the hydrangeas to match the color in the photo (first image) I used as the prompt. Because I’d done some other work in between, Gemini assumed I wanted to keep having a teacup in the image. Which is an issue when it comes to AI generated images - it relies on everything you have already asked to build the image. Pretty scary considering how good the photorealistic images are that Gemini created. If you didn’t know any better I could pass this off as my photograph. I even own the actual teacup in the image.

What does it all mean?

In the in end, the creator of the image owns the copyright to the image, whether its a painting, a photograph, an image manipulated or made with a photo editor or an AI-generated image. The next question becomes - should the use of the tool be called out?

EU AI Act Article 50, which obliges providers and users of generative AI to clearly label synthetic audio, images and video so audiences know what they are seeing, becomes enforceable from August 2026. [Thanks Gasper!] See also Code of Practice on Transparency of AI-Generated Content

What this means in practice is that everything that is produced by AI must be marked as being produced by AI - so I can’t pass of the image of the yellow teapot and weird blue hydrangeas as my own photo. While it’s unlikely that this image could cause harm, the creation/recreation of images of people can be harmful.

But why not photo editors? It’s a good question, and I suspect that as more AI becomes embedded in photo and image software, they’ll have to be marked too.

Over the weekend we were discussing my mother’s declining health and I’d been tasked with finding a photo for her obituary. I’ve not been able to find one of her where she’s smiling. Part of the family suggested I just use AI to manipulate some of the photos to make her younger and smiling. After all, what’s the harm?

Agreed by the time we get to that point my mother won’t be able to argue - but at the same time, there is a huge debate raging about “image-based abuse.” [See Janhavi Kapoor’s comments on the topic.] And since this is a family-friendly substack I won’t go into the whole issue of using images of well known actresses’ faces on the bodies of actresses known for not wearing clothing. Apparently it’s disturbing to both individuals. And it raises the question of where do you draw the line.

So, if I can’t find my mother’s yearbook photo, can I use AI to create one - I am capable of doing so. Will I? Probably not.

Is it funny to take my face and put it on my brother’s body. Is it amusing for my sister-in-law to take a picture of my brother at age ten and combine it with a picture of her at age 10? Well, sort of, I personally find it creepy. There’s no harm done here. But where does the line get drawn on doing harm?

People have faked images for years, photographers have manipulated negatives to show one famous person on the body of someone in a swimsuit or compromising position. The addition of AI as a tool is no different.

The answer returns back to the concept of ethics. There will always be people who are unethical enough to fake, fudge, manipulate and now create fakes. It is going to be up to the world to not promote that practice - no matter how amusing the memes may be….

Yes, it’s a long post, but I hope you enjoyed it. Maybe next time I’ll get into the sustainability issues in using AI to create deepfake images. Or I can delve into the bias built into AI that impacts image generation. Or even explore the ethics of AI generated images further. Let me know what you want to read. In the meantime, to learn more about the Ethics of AI and Images - check out:

• Ethical Pros and Cons of AI Image Generation

• Artificial Intelligence and Images: AI Image Ethical, Legal, & Environmental Issues

• Is it ethical to use generative AI if you can’t tell whether it is right or wrong?

• Ethical Implications of Generative AI in Art and the Media

• Using AI Responsibly For Image Generation: Ethics And Transparency Matter

Let me preface this by saying that I believe the coffee mug you randomly pull out of the cabinet in the morning is a portent to your day.

And yes, the message on this mug is true for me. Which is why a recent debate on whether or not material created by or with the assistance of AI really makes my semantic hackles rise.

I have seen the MLA Handbook come and go. I have gone from typing handwritten notes on a typewriter to now allowing my word processing program to complete words for me automatically. It saves keystrokes – and when you have to type fast, the number of keystrokes matter. Yet, I still use double spaces at the end of sentences. I appreciate spell check. And while it annoys me at times when the word processing program attempts to correct my grammar, it’s mostly because I’ve made a deliberate word choice or phrase and it seems to think I should make it shorter and more concise.

But is this AI? If it is, then we’ve had AI for almost 20 years when Clippy was unleashed on the world. So, here’s where my opinion differs from others. Spellcheck is a tool. Word processing programs are a tool. They may correct your grammar and fix your misspellings, but it’s nothing more than a tool. Spellcheck is pattern recognition and complex Regex (a sequence of characters that defines a pattern) – and while that may sound like AI, it’s not. It was a bunch of programmers hiding in a backroom going through their experiences and asking others about what words they misspelled the most often. (For me it’s initiative and New York. For some reason I try to put an “u” in York.) This is why there are issues with their spell check programming. And why you can interchange words when they are spelled correctly and spell check will not always catch it because they’re correct

Before you had spellcheck in word processing you had to stop, flip open the dictionary and look the word up and then either use the correction tape, or hopefully, just type it. Before word processing fixed your grammar you had to grab a Harbrace College Handbook (of Grammar) and look it up. For style and formatting questions, you grabbed your MLA Handbook, Chicago Manual of Style (Turabian anyone?), or APA Handbook. (Some grammar rules were included in those, but they were mostly style guides. Or in the case of the APA provided guidelines on how to remove bias from your writing, something the original AI needed to be taught.)

Word processing simply replaced the old-school stop and look it up (or ask someone) and allowed you to get something done faster. Is this AI or is this simplification? To me it’s a tool, so that makes it simplification. While in theory it is a form of machine processing – it’s not machine learned – it was hard programmed from the old-school books and manuals. That’s what made it so hard for non-English typists – the original programming was formatted to fix only English rules and style. But not necessarily grammar (the their, there, they’re debate).

Which brings me to what started this diatribe – When do you note if something is AI-assisted or not. While one individual I know asserted that everything he does should be labeled “coffee-assisted” as his brain doesn’t function with coffee – it’s a decent point.

Here are my two cents, for what it’s worth, not withstanding the fact that my background is journalism, library science and multimedia production and design.

Is AI a tool? It could be called that – but given the fact that the point of most LLMs (aka what most people call Generative AI) is to follow a set of instructions to put together new concepts – it transcends the tool bar. Tools only “assist” and while you can use AI to help you fix your writing (grammar, style, etc) its main purpose is to create that information for you. So, this means that if you use AI to create any part of what you are passing off as your own you should note that somewhere. It could be a disclaimer at the bottom, it could be in the middle of your writing, or it could be at the top where you start off saying you had help from AI.

Just like the kid at school that brought the bathroom cold-fusion experiment for the science fair you can look at it and say “their parents probably helped.” It’s usually obvious when something has been enhanced or assisted.

People are smart and for the most part they can tell what been written by AI and what hasn’t. And yes, there are now programs to insert spelling errors, common grammatical errors and the like into AI-generated content to make it sound more human. And I recently had a disagreement with a non-native English speaker who insisted since Copilot wrote the text it was correct. No - it was wonky. We debated for 10 minutes before I just corrected it.

I am terrible at spreadsheets, pivot tables and chart creation – so I love having AI to literally just tell it to create a bubble chart using the data in A1:E26. So yes, people who know me can look at a spreadsheet and know it’s been AI assisted (that and the AI embeds a notation that it was done with AI).

I use AI to summarize large amounts of unstructured data or transcripts – but I always outright state it’s AI-generated. If I write something and I have AI assist me by doing research and writing on my behalf, I note that. But that’s my choice. I’m personally not comfortable with not telling people that AI did something. I take pride in my writing and as a result, I want people to know when I didn’t do it.

I recently took a prompt certification course, and I almost failed an assignment because the AI -grader flagged it for plagiarism. I had a direct quote from
Franklin D. Roosevelt (FDR) in a slide and gave him full credit for it. But since it matched word for word for word to what the AI could find on the internet it was flagged. Before we had AI to flag this for us, we had to guess what might have been plagiarized and look it up (first in the book or library and then on the internet). I’m sad to say that I had a vendor removed as being a trusted vendor because one of the slides I read from them sounded familiar. They had copied a statement (well, several statements) from a news article on the internet and pasted it in as their own analysis. After spending hours going through their decks – it was determined that more than 50% of what was written was provided without a citation (plagiarized) and passed off as their own.

Now, why is this important to a discussion on if it’s AI or Simplification? Because, AI – at its heart – simply reads what it can from the training material (aka Internet) and compiles it back in an order it thinks it makes sense (a very complex Regex). In reality if you don’t check your AI produced content you could be fully plagiarizing and you don’t know if you don’t check it.

Spell check and grammar check just fix what you have written – now if you’ve copied it off the internet, it won’t know, although it will correct errors if they exist. You are not violating any sort of copyright or use guidelines by using word processing or spell check tools. However, using AI is when you get into the grey area of copyright and use. And why understanding how to properly cite your content is paramount. And while Grammarly now has a built in plagiarism checker - it’s built on AI, so you’ll be relying on AI to check AI’s plagiarism.

So, a fun (well if you’re a word geek) party game is to write one word on a card or slip of paper from two standard typing sentences (search pangram) and see how many sentences a team can come up with. I played this game with Gemini and it took about seven prompts before it got it right – and the last prompt included me giving it examples of what I wanted, which it then repeated as its first answer.

When I asked Gemini to give me the total number of logical sentences that could be made from the two sentences it told me (very politely) that it could not be done. I was reminded that “grammar is flexible,” “logic is subjective,” and “combinatorics explode.”[sic] Which is quite true. (For those of you not acquainted with [sic] it means it is quoted exactly as written - which is usually bad grammar or misspellings.)

In reality I believe there will be a day when no one needs to note that AI was used to create something because it will be just as commonplace as word processing. Just as before you could tell when something had been typed and when something had been typed on a computer – for now it’s best to at least somewhere show if AI was used. It’s not a visible difference, unless you suddenly start writing in perfect Mandarin and you’ve never studied it in your life – then it’s easy to tell it was AI generated.

One parting thought – true AI-generated content that’s good enough to fool a human (or other AI) takes a vast amount of prompting – and that in and of itself takes true work and creativity. So, while I may produce AI newsletters sometimes – what you didn’t see was the 40 hours of prompt work that went into them. They’re noted as AI-generated, but only because I feel it’s (personally) morally wrong to pass off another’s work (even if it is AI) as my own.

And that is where the bottom line is – if it’s merely a tool to you then you don’t note it – and you take the risks associated with that. If you believe in giving credit where credit is due – then you do note it. Either way it’s a simplification tool.

Note: None of this was authored by AI, unless you count the brAIn.

Photo Copyright 2026 Elizabeth Roberts - stay tuned for more of my thoughts on AI and photos.