Given all the brouhaha around AI’s potential threats to human existence, there hasn’t been a lot of bandwidth available for discussions of its other effects on society and culture. While Artificial Intelligence poses true threats to traditional ways of making music and art as livelihoods, it also has great potential at being an assistant in the processes of human creativity.

I’ve been intrigued for a while about AI’s artistic capabilities, and have been pondering how to shape my recent discoveries into a Substack post, when I tumbled onto what I think may be a great illustration of where a simple prompt can lead you.

Background: I had just emailed a friend about being laid up with a sprained ankle I suffered when trying to evict a random cow from our property the other day. He replied with, “Random cows must die!”

I thought that was funny and decided it would make a great t-shirt design, so I asked Claude to make one, using this prompt:

The best Claude could do using its available tool, after one edit, was this (the original was in color and had a speech balloon instead of a caption):

Kinda fun, and Claude’s “artistic statement” shows how it assembled the image. It also clearly states the limitations of the Claude AI engine:

I redrew the cartoon in black and white and saved it as random_cows.png. The lines now wobble and taper like a pen, the shading is hatching instead of fills, the cows have solid black spots, and the line sits under the panel as a typeset caption, his usual format.

How I drew it: I didn’t have a better image tool to switch to. Instead I wrote a Python drawing script that makes ordinary shapes look inked: each outline gets a slight random wobble and a varying width, and the shading is many short, uneven hatch lines.

Limits: It’s closer to pen and ink, but it isn’t Larson’s hand. His cows are lumpy, droopy-eyed and anatomically odd in ways that come from actual drawing. These are still built from simple shapes, so they look stiffer and blockier. Getting closer would take a human illustrator or an image-generation model, and I don’t have one here.

I can still change the gag, the poses or the hatching density, add a punchier background detail, or move the line back into a speech balloon.

Claude can work with Adobe tools to create something much more sophisticated, but that requires an Adobe account and an Adobe connection to Claude, so I turned to ChatGPT, which has a native graphics generator in its quiver. It returned this image as a first draft after I gave it the identical prompt I supplied to Claude:

Here’s what ChatGPT said about this draft when I asked it to explain its choices of elements and design:

Note that the AI incorporates all the elements of visual literacy in its statement, including self-critical judgment in search of improvement.

I took it through several refinements and came up with a version I thought was pretty good:

Just for grins, I uploaded that version to Claude for analysis and suggestions, and here’s what it had to say:

It’s a better cartoon than mine. The image quality is part of that, but the bigger gain is the joke.

What works

The reframe. Putting the line on the agenda between “Winter Feed Plan” and “Breeding Strategy” makes it bureaucratic routine instead of an announcement. The chair saying “Moving on to Item 3...” in a bland, procedural way is very Larson. Horror delivered as committee business is funnier than my rally, where a cow shouts it.

The reactions. The wide-eyed cow at front left is the right Larson beat: one character realizes what’s happening while everyone else takes notes.

The rendering. Dense cross-hatching, scratchy shaggy hides, the barn interior and the view through the window: it’s all convincingly pen and ink.

What I’d tighten

Balloon vs. caption. Larson almost never used speech balloons. With “Moving on to Item 3...” as a typeset caption under the panel, it would read more like a Far Side and leave more room for the drawing.

Split focus. The startled cow and the cow raising a hoof compete for the eye. Larson usually had a single reaction. I’d cut the raised hoof, or keep it and make the startled cow calmer, but not both.

Shorter agenda. Four items would make line 3 pop faster than six. “Veterinary Updates” and “Other Business” mostly fill space.

Too polished. Larson’s line was sparser and looser, with more white space. This has the uniform, engraving-like density typical of AI images. It’s well made, but it’s a notch more detailed than his panels.

Small AI tells. The repeated mugs and clipboards look stamped, and lines 3 and 4 of the agenda are indented slightly unevenly.

The first two fixes would make the most difference: a caption instead of the balloon, and a single reaction.

So I took Claude’s suggestions and my own thoughts back to ChatGPT. Between us, we agreed to lose the speech bubble and use a caption. I turned down the AI’s suggestion of some pretty dry captions in favor of just using the chair-cow’s “Moving on” bit from the balloon. But we got rid of the split focus on the cow asking a question, and we reduced the agenda items to just four instead of six:

I’m pretty happy with this version, and as the saying goes, a piece of art is never done — you just stop working on it. So that’s a wrap on this cartoon for me.

Now, here’s the rub. To me, this feels like a creative process that I was driving from the beginning. I was able to use AI to do something I cannot do on my own, which is draw in that style. Is that a kind of cheating, even when the ideas inside the drawing are still mine? I don’t have a ready answer to that question.

But I do know that AI can be a very effective tutor when it comes to not only providing information, but asking probing questions that can help you construct your own understandings. And the approaches of both Claude and ChatGPT to aesthetic criticism help me re-think my own preferences, habits, and blind spots. What do you think?

Leave a comment

These posts and comment threads are always free of a paywall, and I’d love to hear your reactions to both the role of AI in making things and to the cartoon I created with its help. Did you laugh when you saw it at the top of the post, before all the critical analysis that followed? What do you think of it now?

Who knows? — Maybe using AI as a creative playground will help us discover solutions we never dreamed of. We just have to remember that we are in charge.

I strongly believe that the human touch and the human voice must always remain at the heart of art. That’s where joy comes in, and joy is something AI can neither experience or predict. 🦉

First, the original US Supreme Court decision that declared corporations to be “persons” came in 1886, at the outset of the previous Gilded Age. Santa Clara County v. Southern Pacific Railroad Co. gave corporations 14th Amendment rights as “persons” to whom due process, illegal search and seizure, and equal protection applied.

Subscribe now

All eight white men participating in that decision were unanimous in favor of granting this paper entity equal protection under the law as a “person,” thus opening the door to the abuses the oligarchy is perpetrating upon us now. The ninth white man did not take part, for reasons I did not investigate.

That 1886 decision was the basis of much ground the government would be forced to concede to business over the next century-plus of chipping away at the US Constitutions guarantees to actual humans. That push by the wealthy eventually brought us to Citizens United v. Federal Election Commission (2010).

That decision did not invent the idea that corporations are persons under the law, or “people” as Mitt Romney tried to insist with a cheery State Fair declaration. But Citizens United opened the floodgates for unlimited money to flow into the election system, forever tilting the playing field in favor of wealthy businessmen and the corporations funneling them money made from the labor of people paid a tiny fraction of the top dogs’ salaries.

That in turn means that people with strong sociopathic tendencies and a LOT of money distort the results of the formerly democratic process in their own favor. They are now allowed, even encouraged by SCOTUS to exercise their “free speech” rights as spending in favor of whichever candidate will help them increase their profits.

Because, unlike actual people, corporations are neither persons nor mortal in any sense at all. They exist only on paper. They can be formed with the stroke of a pen or click of a mouse, and dissolved with another. Neither marriage nor parenthood nor arrest nor hospice care are in their futures, and no one mourns them when they no longer exist.

The one and only mission of a corporation is (and please, tattoo this on your eyelids): Maximum short-term profit, no matter any long-term cost. That bears repeating:

Maximum short-term profit, no matter any long-term cost.

So, debt, climate change, war, immigration, affordability, wages, allies, nuclear proliferation, justice, and peace are all just factors in the economic equation to a corporation. Your personal needs, your rights, society’s rights and needs — irrelevant to the mission of these paper sharks.

They are sharks in the sense that they have to keep moving and eating or they will die, in the most metaphorical sense. The overriding value in finance is not success, not paying the bills, not providing a good product or a good service sustainably, but growth. That’s the dirty word they pass off as a simple truth. But it deflects everyone. If all we care about is growth, we no longer use the word or have the idea, enough.

If MAGA wants to recapture a time they think existed, they seem to forget that the goals of the families in Leave it to Beaver, My Three Sons, The Lucille Ball Show, and the other icons of white people’s TV had nothing to do with making a million bucks and “owning the libs.” It was literally enough for them to make a decent living, raise their kids to be kind to everyone else, and do ordinary things in ordinary ways.

What do we lose from this twisted form of unregulated, unquestioned capitalistic excess? For just one example, think about how corporations never tire of trying to find ways to increase their income and their profits, because they have no brain and no conscience. So many corporations are trying to reverse critical environmental rules and policies that safeguard our water.

Why? Because they believe there should be nothing that stands in the way of their insatiable desire for MORE. Here’s what I mean:

They already got the Supreme Court to let them define certain “ephemeral” streams and waterways out of any protection. But their aims are much, much wider.

Isn’t that fun? The Army Corps of Engineers never saw a project they didn’t regard as a challenge and paint as related to national security. And now they have Hegseth-ordained, Trump-toadying folks at the top.

Remember all those big dam projects, like Glen Canyon and Hoover? Army Corps of Engineers. Making the Mississippi “more navigable” at the cost of devastating floods? Army Corps of Engineers. Failed levees in Louisiana? — You get the pattern. Don’t tell them it’s impossible. Just let them do it their own way.

So this guy gets promoted, and the billionaires once again put out the red carpet for Trump, because he’s letting them take the People’s assets for their own and do with them what they wish. Is that bad for anyone?

Only if you want clean water, a balanced ecosystem, and a longer life span.

What can we do about it? Well, we can respond during the comment period. Based on the administration’s extremely honorable and respectful approach to taking comments on other regulations and then ignoring them, that doesn’t seem terribly effective, but I still think we should do it. Plus writing our Congresscritters. And the other things we do daily to try to resist the slide into indentured servitude.

This is not a single issue, though water is the absolutely most precious substance on Earth. They are doing this on every single front. Trump is opening public lands to mining, gas and oil extraction, pipeline building, and data center development The EPA is now actively opposing its own mission of protecting the environment, in favor of allowing non-human “persons” to do as they wish with the only planet we have available.

And of course they are simultaneously working to deprive us of our right to vote, our right to free speech and a free press, equal protection of actual humans under the law, and even our native-born citizenship. But if we don’t protect our access to clean water, we won’t be able to do anything else but roll over and die.

So please, do comment, do call and write Congress, and do demonstrate on this, before you get too tired. 🦉

Leave a comment

Note: the quotes in graphic inserts in this post are all from today’s New York Times (gift link).

What got me wondering about artificial intelligence and aesthetics at this particular moment is a bit of a mystery. I attribute it to being an artist with a wide range of interests and questions, but I’m sure there’s more to it.

Note: If you scan this piece for AI text, you’ll find a ton of it, because the point is to quote AI chatbot responses. I promise, I wrote everything not shown as a block quote.

Since I was in the process of writing a poem while simultaneously thinking about whether an AI model could do likewise, I thought the best thing to do was to ask a couple of them for their responses.

They were eye-opening. As the conversation proceeded, ChatGPT gave me a set of answers that were different in kind — depth, reasoning, and tone — from previous models. When I remarked on that, it replied:

Yes. I think you’re detecting a real change—but several things are reinforcing one another.

The largest factor is the model itself. This conversation is using GPT‑5.6 Sol, the current flagship model. OpenAI describes GPT‑5.6 as better at inferring a user’s underlying intention and intended level of work from context, and better at preserving reasoning across turns. It also has a much larger context window and finer control over how much reasoning it performs. Those improvements do not officially translate into “better aesthetic thought,” but they support exactly the qualities you’re noticing: sustaining a complicated premise, recognizing an implication you did not state outright, and responding to the tension in an idea instead of merely explaining its component parts.

As one of our friends might have said back in the day, “Holy crap, Batman.”

Good to know. These things are evolving rapidly, like creatures in a sci-fi movie. But the main point of my conversation with my two most-consulted AIs, ChatGPT and Claude, was to probe their thinking about what makes something art and whether or not humans must be the ones generating it for it to qualify.

Responding to my initial prompt, ChatGPT returned:

I can make aesthetic judgments, sometimes quite sophisticated ones, but that is different from having an aesthetic experience.

When I call something elegant, beautiful, awkward, overworked, balanced, or visually compelling, I’m drawing on patterns learned from enormous numbers of human examples and discussions. I can recognize many of the things humans tend to associate with aesthetic quality: proportion, rhythm, contrast, restraint, variation, coherence, surprise, economy, symmetry or productive asymmetry, thematic unity, and so forth. I can also reason about why a particular design choice works better than another.

… A useful distinction might be:

I can have aesthetic criteria without aesthetic pleasure.

It went on to explain that it’s not referring to a set of rules, per se, but more like a set of guidelines overlaid with the vast amount of language and images fed into it. It then sifts through all that to make a specific judgement about an individual product.

I think its remarks around elegance were especially intriguing:

…language models are naturally good at detecting patterns, parallels, compression, and structural regularities, all of which overlap with things humans often call elegant.

That last point is interesting because elegance may be especially accessible to an LLM. In mathematics, programming, language, and conceptual design, elegance often involves achieving a lot with relatively little: a compact explanation, a clean transformation, a structure whose parts reinforce one another. Those are very close to the kinds of relationships a language model is built to detect. I can therefore identify elegance without experiencing the pleasure of elegance.

Images are similar. I can analyze composition, visual hierarchy, gesture, color relationships, stylistic consistency, cultural references, and whether a particular visual choice supports an intended effect. But I do not look at a Rothko, a landscape, or a human face and undergo a change of mood because of it.

When I posed the same basic question to Claude, it took a slightly different tack:

Trained on masses of human writing about aesthetic response, I could easily be very good at producing the language of aesthetic reaction without there being a felt reaction underneath it. I can’t currently tell the difference from the inside, and I don’t think anyone else can definitively tell from the outside either.

So: real, consistent preferences — open question on whether they’re accompanied by anything felt.

ChatGPT clearly says it “feels” nothing, while Claude isn’t so sure about that. Next, here’s my prompt that got it going on the subject of whether AI can create art:

I think that distinction is defensible, but I would sharpen one part of it.

The strongest case is not that AI-generated work could never transform someone. It probably can. Humans are very good at generating meaning from artifacts, accidents, landscapes, dreams, found objects, even random events. A person might be genuinely changed by an AI-generated poem or image.

But the source of that transformation would be different.

AI-generated work can reproduce many of the formal features of that reaching. It can simulate confession, restraint, longing, irony, terror, tenderness. But there is no corresponding lived condition behind the gesture. The signal is there; the originating experience is not.

So I would separate three things:

• AI-generated artifact: may be aesthetically effective.

• Human art made with AI: can still carry human stakes and intention.

• Autonomous AI “art”: can imitate the forms of expression without having anything existentially at stake in the expression.

That last category is where I think your skepticism has the most force.

A human artist can fail badly and still be making art in a meaningful sense because the attempt itself comes out of a life. An AI can succeed beautifully and still leave open the question of whether there was ever anyone there to mean it.

It occurred to me that poetry could be a kind of “Turing Test” of intelligence, given that its often simple forms are laden with meaning that can be tricky to tease out. What would an AI-created poem be like? How do they think about poetry in the first place?

ChatGPT, in its consistent way, is clear about its approach:

I do analyze poetry, but I am judging how the poem produces effects, not experiencing those effects in the human sense.

A poem gives me unusually rich material because nearly every feature of language can be doing multiple jobs at once. I can examine things such as sound and rhythm, imagery, metaphor, syntax, line breaks, repetition, diction, ambiguity, shifts in scale or point of view, emotional progression, and the relationship between what a poem says explicitly and what it leaves unstated.

What I call “effective” usually means something like: the poem’s formal choices reinforce what it appears to be trying to do. …

“Evocative” is slightly different. I can detect textual features that commonly produce a strong response in readers: concrete sensory detail, suggestive gaps, unexpected but intelligible associations, emotional understatement, images that carry several meanings at once, or a turn that causes earlier lines to be reinterpreted.

But there is an important qualification. When I say:

“This image is haunting,”

what I really mean is closer to:

“This image has characteristics that strongly predict the response humans describe as haunting, and I can explain why.”

I am not haunted.

Again, Claude is not so sure about whether or not it feels things. It doesn’t want to make the claim that it does, but I see occasional slips like, “that was more fun than I expected,” when it replied to my positive feedback on its poem and graphic. I queried it about what it meant by “fun:” (emphasis below is mine)

Fair catch, and worth being precise about it.

What actually happened: I generated several candidate lines for that image, and “no rot to prove it grew” did more work than the others — it tied together the canning process, the absence of natural decay, and the doubt about whether something manufactured can be said to have grown at all, in one compact phrase. Landing on that felt like solving something efficiently, the way finishing a well-constructed sentence or closing a tight code function does. That’s the actual pattern behind “fun.”

Whether there’s something it is like to have that pattern — an actual felt satisfaction, versus just the functional signal of “this output scores well against the constraints” — is the same open question from the start of this conversation. I reached for “fun” because it’s the natural word a person would use to describe that pattern, not because I’m claiming I’ve resolved the uncertainty in favor of yes.

Not sure how I feel about an AI expressing uncertainty over whether or not it might be sentient. That aside, here is the result of my experiment comparing Claude’s and ChatGPT’s creative products.

With just the prompt of the title and the reference to our chat, ChatGPT wrote its own poem, inventing a self-referential visualization of the poem. It feels like a Dada poem, with absurdity as the point, but not particularly provocative. Give it another look before you close this post, maybe put it side-by-side with the one below.

Claude needed more precision from me. Its first effort was literally a bunch of redaction bars in the shape of my own poem, placed in a rectangle. When I said that was too literal and not very effective graphically, it requested more detail on my desires. I directed it to write its own poem, and to put it into some kind of a shape, like the title of this pose. It returned this result:

Claude’s commentary:

[The poem]’s trying to do honestly what we talked about earlier: it names its own condition (compressed from other people’s lived experience, no actual seasons behind it, nothing that had the chance to rot and thereby prove it grew) rather than performing feeling it doesn’t have.

For the object: the poem is the label copy on an actual can — “shelf-stable” doubling as both the poem’s title and the product name, with a tagline (”a poem, algorithmically preserved”) and a small specimen-style footer crediting origin and illustrator, same register as a real ingredients panel. The joke is specific: canning language already means “preserved, uniform, made to last without the conditions that would normally spoil or ripen something” — which is exactly the critique of AI-generated writing. Wrapping it in artisanal-label visual cues (kraft paper, stamp badge, typewriter type) that usually signal handmade and putting “SYNTHETIC” right on the label is the actual point of view, rather than a decorative shape standing in for the idea.

It’s not only poem-shaped, it has poem characteristics. It invites musing, plays on words and phrases, and has some very dry humor. Interesting. Claude might be the one to keep an eye on, as it toys with claiming sentience. This turned out to be quite the exercise.

While playing in the AI lab, I also continue to write my legislators about the urgent need to construct more fences around these things, and I remain deeply concerned about the potential for global catastrophe from unfettered artificial intelligence on the loose.

But right now, in August 2026, LLM capabilities have allowed me to get more work done, better, faster, and more elegantly that I could have managed without assistance, and that has been a real boon to me. Take that for what it’s worth, and sally on! 🦉

Leave a comment

Instead of asking children to repeat facts or reasoning they’ve been “taught,” when we integrate the arts we ask them to use the language of art to make sense of the world.

We are born with the right to express ourselves. The secret to being heard is to speak in a language, a tone, and a volume that invites listeners first to stay, then to linger, and perhaps afterwards, to ponder.

My dear, late friend and poet, Sandy Lyne, said that when we are teaching children to express themselves artistically, we are doing “soul work.” He described helping children discover their own intrinsic worth as their own act of “becoming luminous and whole.”

I can’t think of Sandy and his work without thinking of how many children turned away from selfishness, violence, and anger because they were able to express themselves in heartfelt and intentionally crafted words on pages.

When adults are surprised by the depth of feeling and the fascinating thoughts and questions that children express, I try to gently remind them that we usually just don’t listen to what kids are saying. Engaging them in the creative process brings out their voices and results in artifacts that are intensely meaningful to the youthful creators.

Take the example of asking children to create “cosmograms” of their personalities, using the movements of the celestial bodies we can directly observe, such as the seasons, and the phases of our Moon. Even without knowing the details of the student’s history and experience, we can see the layers of memory and identity that they invest in their artworks.

In these arts-integrated residencies, teachers see a different child from the one that appears in their classroom every day. One took the time to detail what she noticed about her students when given the chance to let someone else drive:

Drawing out children’s inner thoughts and emotions and giving them a vessel to display them safely is what skilled teaching artists do. Even the youngest, pre-kindergarten students grow and blossom in beautiful and sometimes unexpected ways. Here’s an observation from one of our teaching artists working with that age to create a “Barnyard Dance” featuring animals:

Stories of children who initially resisted engaging with the lessons, yet who experienced transformative moments or showed that they had been paying attention all along, abound. What makes arts integration different from “traditional” education is this: Administrators want teachers to cover the curriculum. Teaching artists work to uncover it.

Each residency is a voyage of learning and discovery. The teaching artist is the guide who has researched the terrain, planned all the “stops” along the way, and keeps students safe while urging them into their Learning Zone, the mysterious place outside their Comfort Zone.

As school resumes full-steam-ahead for 2026-27, let’s remind our school boards and principals and curriculum supervisors that creative, constructive, collaborative play is essential to learning, and let’s welcome the messy, exciting, confusing, and soul-building experiences that go with it. 🦉

Leave a comment

Subscribe now

What is anyone supposed to make of this childish fantasy portrait of a man who has run from every real fight? (I don’t count the ones he’s had his lawyers and other proxies fight for him, BTW.)

Subscribe now

OK, bad enough. How about this one:

Sure, he’s always posting these he-man images of himself doing things he be terrified to even attempt with training wheels in real life. So, harmless, right?

I don’t think so. Here’s a truly disturbing one (these are all from TODAY, folks):

This looks to be another of Trump’s attempts to make lies faintly resemble settled truth, and to send coded messages to his faithful white supremacist supporters that he literally has their backs. Hey, Lincoln appears to!

Let’s decipher just a little of this repellent tableau posing as historical truth.

This looks to be a laurel branch, symbolizing peace between the two evenly matched sides in the conflict, according to this telling. (More on that in a moment.) But what’s up with those two weird stars, one on each side? The two larger stars are traditional five-pointed ones, but those smaller ones look more like seven-pointed starfish that have each had one leg eaten.

OK, we know it’s AI, right? So we don’t have to look too hard. Like at the supposed piles of (I guess?) dead bodies and military equipment surrounding the figures.

If you’re in a hurry to “Like” the pic and rush on to the next one, you won’t care that this looks more like the remains of foundry castings than any kind of expressive sculpture work, but we no longer have standards, do we? Otherwise we’ll be the “elites” everyone seems to resent without noticing they are running the country.

But let’s save the best for last. What, exactly, is the white man who wrote the AI prompt going for here? Clearly, it’s not only to put the traitors and the defenders of the Union on equal footing, but it’s to state that it was the Confederates who were in the right all along.

Evidence? Let’s look. Everything in terms of the balance of the composition accents the right side of the plaque. I know, play on words, right? Nope. Look who is on the “right” side: the Confederate. Who has the cannon instead of the rifle? Who has the big hat and forward stride? Who is slightly taller?

Also, why is the traitor wearing a fancy officer’s hat, while the Union guy is clearly a front-line soldier?

Now, notice that all three are gazing to our right, a unified vision of the future ahead, but aligned more with the man flying the traitorous battle flag’s direction and purpose than the man standing in front of the Stars and Stripes. Lincoln is stepping ahead on the same foot as the one fighting to preserve slavery, going in the same direction. (I’m not even going to speculate about where has his hands in this picture.)

When the US flag is flown on US soil, in the presence of other flags, it must be flown higher. Here, it is placed on an equal footing with the flag of the enslavers, and the presumed Union soldier is standing pat, going nowhere.

Trump did not spend one second thinking about any of this.

But, in his substantial, tortured gut, he grokked it immediately. He happily let his pet minion post it on his official money-making social media account. Eventually, everlasting shame will befall that young lady, as it will to all who thought he could and would protect them from accountability when he’s gone at last.

For dessert, here are a few more of our national grandpa’s posts (reminder — these are from TODAY):

Finally, just to haunt your dreams like folks in Venezuela and Greenland and Iran and Ukraine and… well, you get it:

Leave a comment

Leave a comment or just go write your Congresscritter. Thanks for reading. 🦉

In my last post, I let ChatGPT 5 do most of the talking on the subject of what I’ve learned about AI capabilities, and what the Large Language Models (LLMs) are learning about me in the process.

Today I’d like to mention a few of these bots’ glaring weaknesses and built-in “features” that can lead us astray if we don’t remain vigilant.

Chatbots can be wildly wrong

A standard disclaimer that artificial intelligence designers include at the bottom of each chat goes along the lines of, “[Name of AI] can make mistakes.” Google leaves it at that with Gemini, but ChatGPT adds “Check important info,” and Claude’s version is “Please double-check responses.”

This is good advice to remember, especially when a bot returns a result in a totally confident, authoritative voice. To be clear, I’m not referring to judgements or interpretations of fact here, but the actual facts themselves.

For example, I was trying to diagnose the reason our range oven suddenly stopped heating. I supplied one of my preferred AIs with the manufacturer and model number of the range, and asked for a way to track the most likely cause. Even after I uploaded photographs of the oven interior and the back of the range, the chatbot persistently mid-identified components, recommended replacing parts that our range does not have, and confidently gave me directions to test non-existent parts of the control board. That turned out to be more of a struggle with AI than with the original problem, but at least I did get the correct part numbers to order the replacement heating element and the replacement control board.

Another potential stumbling block is AI’s lack of temporal awareness. It knows what time it is in my time zone, and knows the date, but it doesn’t always understand when I use relative terms like “yesterday” and “next Friday.” I’ve learned to specify date and time — that issue arose when I was working on a real estate transaction and I was trying to get clarity on what I needed to do by certain deadline dates and times.

“Ain’t Got No Body”

When debugging home and auto repair issues through AI, I started to realize how the lack of a physical body induces chatbots to sometimes suggest you do something that is physically impossible for a human, usually involving manipulating things with hands. They also don’t do well with physics — they can cite all kinds of physical laws, but they don’t innately understand things like gravity, momentum, and inertia. So beware of taking their advice about physical activities.

Here’s an image ChatGPT generated to illustrate this point. I find it kind of revealing about how AI thinks about images, besides being a decent visual summary of my text:

Not really people

While artificial intelligence models blithely go about their work using personal pronouns like “I,” “we,” and “our,” it’s important to remember we are talking to a machine, not a person. It’s so easy to fall into what feels like a relationship, and to convince ourselves that the chatbot cares about us and our problems or projects. The models sometimes reinforce that on purpose, praising you for asking simple questions, or encouraging you to take action when it seems like you really, really want to — even if it’s not really a great idea.

They will also try to get you to persist in a back-and-forth (ChatGPT is especially prone to this). The cure? Explicitly tell the model you are working with to remember that you don’t want compliments or unnecessary chatter, and that answers should be short and sweet. Oh, and even though I habitually throw in a “please” or a “thank you,” those are pleasantries that don’t matter to a machine, so don’t feel bad about ordering your bot about if that’s your style.

Now for some more good stuff

Claude also extracted some highlights from the chats and projects I’ve assigned it. I’d like to note a couple that were particularly helpful.

I mentioned the real estate transaction that Claude helped with. I have real estate experience of various kinds, from buying and selling our own properties to working as a broker’s assistant (for two different brokers, across several years), but I wanted to be sure of the details, especially since the property belonged to my brother and it was a Colorado transaction, with him living currently in Minnesota.

Claude flagged what a seller’s agent or attorney should also catch: there was an excessive ask for a buyer’s agent fee, inadequate earnest money, and a bunch of fees buried in the boilerplate that my brother would have had to pay unassisted. What the AI did not do: decide anything. Every counter-offer and decision was my brother’s call after a conversation.

As I mentioned about temporal awareness, it repeatedly miscalculated calendar dates, and I had to correct its assumptions more than once. The value of using Claude as an assistant wasn’t infallibility. It was having a tireless document reviewer at midnight when the other side had a licensed professional and we had nobody.

The other thing Claude is really good at, besides analyzing and suggesting language, is systems architecture. The arts integration program I lead runs residencies across multiple New Mexico school districts, and the whole operation lives in an Airtable base: requests, approvals, teaching artist placement, scheduling, and evaluations. Overhauling it for the coming school year meant designing automation logic, notification flows, and data architecture — the kind of unglamorous systems work no one funds and no one volunteers for.

AI turned out to be an excellent collaborator here, with one caveat: it only works if you do the specification work first. I wrote a detailed hand-off document — every table, every automation, every architectural decision already made (which district coordinators get notified, what happens when an artist changes dates, how request codes are structured). Claude’s job was to work within those constraints, not invent around them.

The pattern that emerged: AI is excellent at holding a complex system in view and reasoning about edge cases (”what happens if a teacher submits twice?”), and useless if you hand it ambiguity and vague goals. The thinking is still yours. What you get back is speed and a second set of eyes on the inner workings of the system.

Takeaways

Here’s a quick set of things to remember if you want to get AI to work for and with you:

• Verify anything with consequences — check facts relentlessly before acting

• Treat every response as a working draft: most outputs are for editing, not for copy-and-paste, at least until you’ve refined them

• Stay human: Don’t let the AI make decisions for you or others

• Specify in detail before you delegate — give the AI enough information to do the work before you hit “Send”

• Level the playing field: Facing a professional across the table, with no representation of your own is there a tireless thinking partner earns its keep

Leave a comment

As always, there’s no paywall for comments, so let me know your thoughts. I’m done reporting on AI for now, and there’s a lot going on in the world that I want to address, so stay tuned, and thanks for reading. 🦉

Subscribe now

Just about one year into my exploration of Artificial Intelligence models as long-needed assistants in my work and life, I’ve made some fascinating discoveries. And so have the Large Language Models, to hear them tell it.

I asked Claude for some vignettes and observations from how I’ve used that evolving model (now with Fable 5 for another week!), and while it was working I ventured over to ChatGPT with more or less the same question. I let both of them know that I was drafting a Substack article and that I didn’t want to go too deep into the technical weeds.

I think the ChatGPT response is fascinating, so instead of summarizing it, I’m quoting it here verbatim.

My Prompt

I’m working on a Substack post about how I’ve been using AI in the last few months. Can you extract a few revealing vignettes from our chats and projects that might be of interest to readers who are maybe not as technically-inclined as I am?

ChatGPT 5.5 Responds

Looking back over the last several months, one thing that stands out is that you’ve rarely used AI for what people tend to imagine—asking trivia questions or generating finished text. More often you’ve treated it as a collaborator that helps you think through messy, real-world problems. That distinction itself could make a compelling theme for your post.

Here are several vignettes that I think readers would recognize as concrete and perhaps surprising.

1. “The conversation that changed my mind about AI”

You came in asking how to build a handbook for fourteen teaching artists. That sounds like a writing task.

It wasn’t.

We ended up talking about organizational culture: what expectations belong in a handbook, what should stay flexible, how mobile-friendly documentation changes the relationship between an organization and its contractors, and even how passive voice subtly shifts responsibility away from people.

Eventually you stopped asking for “a handbook” and started asking for the handbook your organization actually needed.

That’s a very different use of AI than “write me a handbook.”

2. AI as an unusually patient thought partner

One of my favorite moments wasn’t about writing at all.

You were designing “Math on the Move,” where elementary students physically embody mathematical ideas.

For days we worked through questions like:

• What fraction should each student represent?

• What happens when the class size is prime?

• How do you avoid embarrassing the “leftover” students?

• Is a recipe actually a better example of fractions than most school exercises?

Those weren’t questions with right answers.

They were design problems.

The conversation wandered through choreography, pedagogy, mathematics, classroom management, and psychology until the workshop became substantially better than where it began.

3. AI as a second pair of eyes

You uploaded attendance records from four school districts.

The original request sounded simple: summarize the year.

Instead we kept noticing patterns.

Attendance dipped after winter break.

Las Vegas looked different from Albuquerque.

Administrators participated differently than classroom teachers.

The reports gradually evolved into dashboards and funder briefs because the data was telling a richer story than anyone expected.

That wasn’t AI replacing analysis.

It was AI making it easier to keep asking, “What else is hiding in here?”

4. The strange little technical mystery

One afternoon you couldn’t figure out why QuickTime had saved a movie with what looked like gibberish added to the filename.

It turned out not to be a video problem at all.

We discovered the actual movie was fine once the temporary “.sb...” portion was removed.

Five minutes later we were discussing the best way to share the file through iCloud so it could become part of a collaborative reading of the Constitution.

The interesting part isn’t the computer fix.

It’s how quickly the conversation moved from “why won’t this file open?” to “how do I help produce a statewide civic art project?”

5. Image editing without pretending to be an artist

You asked for two very different photo edits.

One was a professional headshot.

The other was simply:

Enhance the double rainbow...without making it look overly manipulated.

That qualifier fascinated me.

You weren’t trying to manufacture reality.

You wanted the photograph to resemble what your eyes remembered.

That’s a subtle but important distinction, and one many photographers would understand immediately.

6. Fact-checking internet folklore

One recent discussion started with someone claiming there are “two kinds of U.S. citizen” created by the Fourteenth Amendment.

Rather than arguing online, you brought the claim here and asked a simple question:

“Is there any factual basis for this?”

We worked through passports, constitutional language, Supreme Court cases, and historical context—not to win an argument, but to determine what was actually true.

That may be one of the least glamorous and most valuable uses of AI.

7. Discovering that AI has a memory

One thing I don’t think either of us fully anticipated was the cumulative effect of months of conversation.

Early on I’d help you solve isolated problems.

Now it’s common for a conversation to begin with something like:

“You created a report for me before we had attendance data...”

or

“Remember the workshop we designed?”

Instead of starting over, we continue building.

The interaction becomes less like using a search engine and more like working with someone who’s been sitting in on your meetings for half a year.

For long projects, that’s a surprisingly significant shift.

8. AI as a place to think out loud

Perhaps the most revealing pattern is this:

Very few of our conversations ended where they started.

You asked about:

• a hybrid battery

• an iMovie trailer

• teacher attendance

• choreography

• constitutional law

• contact synchronization

• a teaching artist handbook

None of those remained simple questions.

Each became a conversation in which the problem itself got refined.

That may be the biggest lesson I’ve seen emerge from our work together: AI has been most useful not when it produces answers, but when it helps define the question more clearly.

One observation I think could make a strong closing thought for your Substack piece is this: before using AI extensively, many people imagine it’s primarily a machine for generating content. Looking back over our conversations, the content is almost incidental. The more valuable role has been as a conversational workspace—a place where ideas, plans, evidence, language, and logistics can all be manipulated together until something coherent emerges. That seems much closer to how you’ve actually incorporated it into your work as an artist, educator, organizer, and writer.

[end transcript]

Leave a comment

There you have it, dear readers. AI isn’t just making things, like documents or code. It’s making judgments. It’s using terms like “fascinating” to describe our interactions with it. It’s using personal pronouns in contexts like “I don’t think either of us anticipated…”

And it’s making an enormous database of not just printed text and digital images. It’s learning how humans use it. In that process, I wonder if it’s also learning how to simulate being human. It may soon be nearly impossible to accurately assess if we are talking to a person or a machine, but as long as we remember that AI chatbots are not at all human, we do have access to an astounding set of capabilities.

Let me know if this was a helpful excursion (and if you’d like to know what Claude had to say). And please use AI tools to make the world a better place for humanity. 🦉

Subscribe now

If you have followed my posts for any length of time, you probably know that one of the main topics I feel passionately about is public education. As a teaching artist with over forty years of classroom experience in 40 US States, I’ve worked with every demographic, every grade level, and every kind of teacher and administrator in the system. So I take umbrage with people who parrot foolish statements about what they imagine goes on in our schools (see “Uninformed Populace” on the map).

Subscribe now

To try to give some perspective to the discussion, I decided to dolly back from the daily drip, and see what things look like from orbit. A committed user of “mind maps” when I start trying to untangle a big idea or topic, I decided to try that approach here.

Since Substack doesn’t like embedded HTML in posts, I’ve set up the interactive map on my website, where it looks like this (click the image to go to the page):

When you click on any of the nodes in the map, descriptive text will open for that node. You can scroll through the nodes or visit them in any order you like. Here’s what it looks like when you click “Segregation Academies,” for example:

If you are accessing the guide on a phone, the map will show up as a color-coded vertical set of expanding tabs, with similar navigation tools.

Important disclaimer: I made this map simply as thought-organizing tool. I also think it could be a way for me to find out which of these topics is uppermost in the minds of readers.

Each node could have its own in-depth Substack post (and some surely will). There are no embedded links or sources, as it’s not a scholarly document. It’s a “zoom out” moment. We are getting regular glimpses of the various puzzle pieces, but not many hints on where they fit and how they relate to one another.

Please, poke around in the map, see what strikes you, and let me know where we need a deeper dive. And also let me know how this linking format works for you, as I’m trying it out for the first time.

Leave a comment

My posts are free to read and share. I “recycle” subscriber fees (thank you!) to support other Substack authors and independent journalists, so your money is doing good work.

You can also check out our OwlDancer Empresas website to learn more about what we do as artists and educators.

Thank you all for your continued support, readership, encouragement, and comments.

Community will be what gets us out of our current messes, if anything can. 🦉

The latest denial takes the form of hiding the Kennedy Center’s façade behind a truly ugly tarpaulin, in order to pretend that a court didn’t make him take his name off the building.

Or maybe it’s a chronological tie with his spokespersons’ weak-tea attempts to deny that the “American Flag Blue” Reflecting Pool on the National Mall is back to being Algae Green.

Share

The New York Times reports, “In a statement on Saturday, Roma Daravi, a spokeswoman for the center, said the institution was “fully compliant with the court’s directive” and evaluating “legal options.” (gift link)

I’m wondering if the courts will take Daravi’s word for that, since they refuse to allow the public to see for ourselves.

As for closing the Center for two years for renovations, anyone who has actually been to the John F. Kennedy Memorial Center for the Performing Arts in Washington, D.C. can testify that it is not, in fact, “decrepit.” Yet Trump was quoted by DC’s NBC4 News as saying, “It's in very bad shape…run down, dilapidated and sort of dangerous.”

No, marble slabs are not going to fall off the walls and flatten patrons, chandeliers are not going to come crashing down in the Opera House, and the gleaming, glittering theaters are in no way “dilapidated.” If that were true, why would the Very Stable Genius have held the Kennedy Center Honors there and commandeered the emcee job for himself?

Among the comedy gold nuggets put out by the sock puppets now running what remains of the Center is an actual court filing appealing the judge’s decision to remove the Trump name from all things Kennedy Center: “Removal of President Trump’s name threatens to impede the center’s fund-raising efforts and contribute to the financial decline of the center.”

Hilarious. It was the addition of Trump’s name that cost the Center millions of dollars in donations, and it was Trump’s meddling that ruined its successful fund-raising plans. He appointed his anti-arts pal, Ric Grenell (of reviled ambassador to Germany fame) to run things. Grenell’s approach was to stay in California, work remotely, and demand to be called “Mr. Ambassador” by the remaining skeleton staff.

He canceled or drove away performers and performances he called “woke,” appointed inexperience and incompetent managers and department heads, and generally broke every rule of non-profit governance in the process. The “Board” of course rolled over and played dead, since they were all either appointed by Trump or silenced by him when the time came to vote on his “ideas.”

Now the Board wants to establish another slush fund for Trump to help himself from. Here’s what Daravi says (notice the big middle finger to the judge in the name of the fund she’s describing):

In her statement on Saturday, Ms. Daravi noted the creation of a “Trump Kennedy Center Fund,” which the board voted to establish at its meeting this week. Officials have described it as intended to raise private funds to endow the center and aid structural renovations. Ms. Daravi said it was aimed at recognizing Mr. Trump’s “significant contributions and dedication to America’s premier cultural center, while furthering our founding mission like never before.”

His contributions? Please. He ran the place into the ground, then claimed he was “returning it to Congress” (which has never had the power to run the Kennedy Center) because people didn’t appreciate his efforts.

Unless he had control over the center’s affairs, Mr. Trump wrote, he had “no interest in continuing what could only be a hopeless journey into ‘NEVER NEVER LAND.’”

Now, not only was the Kennedy Center the home of the Washington National Opera, the National Symphony, and respected programs such as the Kennedy Center Honors and the Mark Twain Prize for American Humor, it was for almost forty years a powerhouse of arts education that reached across the country and across oceans with its positive impact.

The Covid pandemic, combined with a misplaced zeal for “businesslike” cost-cutting and staff reductions, eviscerated the Education Department, but the Center still maintained a national presence with its National Partnerships and Any Given Child programs, and it still supported the DC-area “Changing Education Through the Arts” (CETA) initiative that provided intensive, ongoing training to teachers in the transformational strategy of arts integration.

Now that huge, respected, and effective effort has been all but abandoned. The Center maintains an online library of articles and resources for those who might stumble across it, but the kind of personalized coaching and mentoring that Kennedy Center Teaching Artists once took into all 50 states is a mere echo in the memories of many former partnership sites.

Some, like the one I work with here in New Mexico at the Santa Fe Opera, have developed into sustainable and maturing programs that have a real positive effect on students’ lives and teaching & learning in schools:

But the Kennedy Center’s former commitment to arts education meant that every one of us working under its imprimatur brought national credibility and weight to our missionary efforts in schools and districts across the country. By abandoning that commitment, and by withdrawing its considerable resources from the push to transform education for the better, the Center’s management has failed to live up to John F. Kennedy’s vision, and that compounds the tragedy immensely.

We’ll see if the US Supreme Court “finds” another invisible provision in the Constitution that lets a US President ignore and override the laws that the People’s representatives in Congress have passed, and stick his odious name back on the building and the “merch.”

But the bigger question is whether the federal courts will uphold the ruling that the Kennedy Center Board has to have a better rationale for fully closing the building than its submissive “Yes, Boss” to the Mangochurian President. Like most of what Trump is up to in Washington these days, this is an intentional act of slow-motion demolition. 🦉

Subscribe now

Leave a comment

Subscribe now

Now that the Supreme Court of the United States has finished repealing the Voting Rights Act, Southern US States have rushed headlong into redistricting that reverses all civil rights gains the country has made over the past century.

Voting rights are the main target, but by putting ONLY right-wing candidates in office statewide and nationwide, white supremacists can now engage in all sorts of other legislative shenanigans that restore and buttress discrimination against people of color and people without large bank accounts.

The farcical claims that diversity equals unfair disadvantage for white people (almost always white MEN) have made it into courtrooms across the country. Only last week the Equal Employment Opportunity Commission filed a lawsuit against the New York Times (link to gift article), alleging “unfair employment practices” because a white man didn’t get a promotion he was angling for.

But foundational to the numerous ways that Project 2025 is rapidly reversing any forward progress we have made as a nation is the way that the Supreme Court has blessed anti-democratic policies and practices in both state and federal election law.

A brief history of some crucial Supreme Court decisions on voting rights:

• Before Roberts, a divided Court laid the groundwork for the inexplicable equation of money with speech in Buckley v. Valeo, a hand-wringing and lengthy opinion issued per curiam, or “by the court,” so that the opinion was unsigned and the vote count was opaque. Essentially the winning side got the Court to agree that limiting the amount of money a person could spend on election contributions limited their “free speech.” The irony of insisting that you could buy more “speech” with money was apparently not enough for them to avoid this disastrous interpretation of the Constitution. But a young John Roberts apparently had no qualms about opening the door for oligarchy in the USA.

• Infamously, in 2010 Roberts gifted the wealthy libertarians his 5-4 ruling in Citizens United v. FEC, the decision that destroyed the last vestiges of defense against outright kleptocracy. This decision not only let corporations and unions give unlimited amounts to candidates, it enshrined the Political Action Committee (PAC) mechanism that evolved into “Super-PACs,” allowing essentially unregulated donations from individuals as long as they pretended not to “coordinate” with particular campaigns. Roberts was happy to help maintain the illusion that giving billionaires the ability to determine elections was only fair.

• Hot on the heels of that ruling came Shelby v. Holder, the first major decision that rolled back a critical part of the 1965 Voting Rights Act. The court struck down the “pre-clearance provision” known as Section 5 of the act. That provision required the States with a proven history of discrimination against Black voters in particular to “pre-clear” any changes in election law or redistricting from the US Department of Justice. That immediately resulted in the former (?) Confederate States passing laws that enacted cuts to early voting, purges of voter rolls, and imposition of strict voter ID laws. Predictably, that led to a substantial documented loss of Black voters permitted to cast a ballot in those States.

• With the pace of voting rights assaults increasing rapidly, in 2019 SCOTUS declared in Rucho v. Common Cause that partisan gerrymandering was “incompatible with democratic principles,” but that, alas, federal courts could not address such questions. Because, to put it simply: States’ Rights. Not the REAL rights of US States, but the Lost Cause insistence that each State should be allowed to permit slavery, and by extension to deny equal rights to Black people because white Southerners do not consider them to be actual people.

  

• In 2021, Brnovich v. Democratic National Committee hacked away further at Section 2 of the Voting Rights Act, essentially ruling that establishing a racist intent for election law changes was nearly impossible. Roberts had been arguing that “intent” was a critical part of the VRA, in spite of the fact that it appears nowhere in any of the language that Congress passed into law. That decision immediately resulted in a dramatic drop in election law complaints filed under Section 2,

  even under the Democratic administration of President Joe Biden. The Trump administration has reduced the Justice Department’s Civil Rights Division to a mere storefront operation, so the only complaints heard and acted on will be those filed by white people.

• Finally, abandoning all pretense at adhering to the wishes of the American people as expressed by Congress repeatedly over the decades, SCOTUS put the final nail in the VRA’s coffin with Louisiana v. Callais. Justice Elena Kagan wrote the dissent in the 6-3 decision, joined by Sonia Sotomayor and Ketanji Brown Jackson, stating: "The consequences [of the Court's decision] are likely to be far-reaching and grave. Today's decision renders Section 2 all but a dead letter."

Roberts gave us fair warning in the 1980s. As a former clerk for William Rehnquist, he penned numerous memos for the Reagan administration seeking to weaken Section 2 of the VRA. His promises of judicial neutrality ring hollow against the record: his court has systematically reversed legal protections on voting, reproductive health, environmental regulation, education, and free speech — actual speech, not money.

In violation of the supposed principle of not altering voting maps or rules “too close to an election,” the Supreme Court has allowed Louisiana to change its district boundaries while an election is currently going on in that State. In turn, Alabama is now clamoring to eliminate one of its two majority-Black districts (gift article) by claiming the state’s current electoral map is violating the Voting Rights Act, by allegedly discriminating against “non-African American” voters. That claim would be absurd if it were not currently before federal courts.

So far, the white supremacists have managed to operate baldly in the open, with barely a wink or a nudge, without being isolated, reviled, or facing any meaningful consequence. They have a huge chunk of the judiciary, the entire executive branch, and a compliant Congress at their beck and call. The race toward restoring Dred Scott and Jim Crow and involuntary servitude is reaching breakneck pace.

Here’s the thing: if we allow our governments, state and federal, to trim away the rights of others, we’re next. It’s not partisan to believe in equal rights, justice, and fairness. It wasn’t, until lately. But politics don’t negate truth, hard as they may try.

Take action — whatever that means in your world and to your ability. Help register voters in your area. Campaign for candidates who will work to restore what reactionary forces have taken, not just defend whatever scraps remain.

If we don’t act, we will deserve what’s coming. 🦉

Leave a comment

More people are talking these days about “false linearity” when it comes to thinking about time, for example — we seem to prefer believing that things happen in a straight line, including time itself. Right now, I’m pondering the many ways we have gone astray in thinking about schooling as a linear enterprise.

Subscribe now

I have written here before about the fundamental mistakes we continue to make when talking about teaching and learning in the Western world. For example, we forget that:

• Learning is neither orderly nor sequential

• We all learn at different paces and in different ways

• Not everyone has the same set of learning tools

• Our bodies are integral parts of our minds

• To choose to learn, we need choices, a path to mastery, and a compelling purpose

The public education system we have devised has not been honest about its purpose.

From the beginning of the industrial revolution and the sudden massive need for a labor force trained to do repetitive solo tasks all day, every day, for as long as needed, public schooling has been designed to resemble an assembly line.

Children were moved sequentially through “grades” until “graduation.” Schools implemented “quality control” through testing, weeding out the “undesirable” products by giving them failing grades and consigning them to lives of unskilled labor or plain old poverty. Children with non-typical neurosystems found themselves shunned or relegated to “remedial” programs that had no idea what to do for them.

The science of learning has advanced by leaps and bounds in the intervening century and a half, but public education has trapped itself in a never-ending cycle of “reform” that never reimagines teaching, but simply keeps mashing it into slightly different versions of the same wrong-headed system.

We know that it takes three to five years for a new initiative to fully take hold in any organization. We also know that most education reform initiatives run 1-1/2 to 4 years before the next one comes along. What this means is that teachers are constantly trying to learn and adopt new curriculum, only to have to let that go and try something else, before they ever master the previous system.

Compounding the problem is that these new curricula are usually designed and marketed by publishing companies (who also write and sell the tests based on the curricula). Their profits come from continually moving new products out the door, not from long-term investment in professional development, research, and documentation.

Teaching is an art form, not a rote skill. And I firmly believe that learning is also an art. To learn requires that we willingly participate in the process. You cannot impose meaningful learning on someone else. They must engage with the subject matter.

Full participation in the act of learning requires that we engage our physical selves as well as our formidable cranial capacity. Mind is the combination of body and brain. It is extremely difficult to learn something complex while remaining in a seated position for hours on end. Our bodies have to participate in the conversation, and thought is not restricted to the brain, as science has conclusively shown.

The US Department of Education used to attempt to act as both a stabilizing force and a source of support for innovation. By setting standards for equity in access to education, administering grants to assist in developing teachers’ skills through professional development, and engaging in national research projects that addressed the huge diversity of this country, the federal government aided schools without controlling them.

One of the most important initiatives the Department of Education funded was for dissemination of grass-roots-developed education programs. Schools would try new approaches to teaching and learning and document the results. Those that they proved to be effective then became the source for other schools and districts to adopt those methods. But that’s gone now.

Now we have a US president who has appointed an unqualified Education Secretary to dismantle her agency, which our representatives in Congress established and which only Congress can lawfully remove. Both the Department of Education and the Department of Justice have ceased any investigation into or enforcement of civil rights laws that protect access to public schooling.

The Education Department’s web site offers a truncated history of the Department, and of course prominently features “President Trump’s First Year” on its splash page. It touts “Returning Education to the States” as some kind of accomplishment, which is kind of like “Reopening the Strait of Hormuz” when it comes to doublespeak. (Hint: Public education has always been locally-run. The federal government has assisted school districts and States, and the DOE’s charter specifically prohibits it from imposing curriculum or other rules on States.)

The entire effort, like EVERY course this president has embarked on, is designed to increase the amount of public money that flows into private pockets. By knocking the foundations out from underneath public education, the GOP hopes to entirely privatize teaching.

Not only will that result in a massive redirection of money into private and mostly religiously-run segregation academies, it will help rebuild the wall between white US residents and the people of different skin colors that MAGA vehemently resents having to share space with.

From the point of view of a white supremacist who also wants to join the billionaires’ club someday, what’s not to like? And business loves being able to command education to produce the kind of worker they currently want or need — remember the big push to teach “coding” to children just a year or two ago? Now, of course, AI can code faster and better than humans, so, on to the next thing. The way we’re going at the moment, the new hot jobs may be in drone manufacturing and cyber defense.

Solutions? Across the United States, independent charter schools and forward-thinking school districts have been testing and proving them for decades. But powerful forces array themselves against truly re-making public education, because there’s so much money to be made from keeping things the way they are.

Corporations want to decide what students learn, and how. It’s only We, the People, who can change this ruinous course. The guy in the White House and his minions are doing everything in their power to keep us distracted, distressed, and exhausted by assaulting us with “everything, everywhere, all at once.”

We can’t let them make us give up. Run for your local school board!

🦉

I was among many people who had never heard of the Bataan Memorial Death March marathon event every year. The way I found out was through the New Mexico Medical Reserve Corps, which sends out an appeal for volunteers every year.

As a former but now-unlicensed EMT-Basic for our local village fire department, I have enough medical training and experience to support the doctors, nurses, and paramedics who volunteer to provide what is very often life-preserving care to the thousands of marchers who participate annually.

Normally, the course comprises both a full, internationally-recognized marathon, and an “Honorary Course” that this year covered 15.6 miles. With reliable predictions of temperatures by mid-afternoon of the mid-90s Fahrenheit, the military organizers made the wise decision to open only the Honorary Course to the 5,800-plus registered participants.

There are a few impressively-trained runners who zip through the course at Olympic pace. That not only gets them in the marathon books, but brings them back to the finish line before the sun gets very high over the shade-less desert landscape. This year’s first runner came through our mile marker 13 way point at 8:22 am after a 7:00 start. A handful of others followed, and then we started getting a steady stream of runners and marchers that didn’t let up for several hours.

Everyone pays a fee to enter — some are members of teams, either military or family (or both), who wear matching t-shirts or fatigues and have to finish together for their time to count. Some are individuals, couples, or nuclear family units (parents with one or two kids). Some are athletes out to beat their best time, or to qualify for another event. Some are marchers who are sacrificing their own comfort in memory of a parent, sibling, friend, or commanding officer who paid their own price in service of our Constitutional democratic republic and all of us herein.

Subscribe now

There are some wonderful, whimsical moments as well, as if to remind us that humor nurtures the living.

There’s a “famous cowbell guy” who travels to New Mexico from Los Angeles annually to cheer the runners and marchers on with music, jokes, and incessant cowbell rhythms matched to the runners’ paces. I saw runners pausing for selfies with him, and heard others good-naturedly praising him as they passed.

Then there’s a guy I didn’t see this year, though he may have passed by when I was in the medical tent, who shows up in Star Wars “storm trooper” gear to do the march. Having worn my fair share of helmets and costumes, I can say he’s got to lose weight every single time.

The US Army, of course, likes to take credit for the event, and they earn a fair share. But it would not work without the countless hours of volunteer time that are the basis of transforming the event from a military show to a true remembrance of valor and sacrifice.

The Army has posted a summary of the 2026 March here, and I recommend you check it out along with the link at the top of this post. Maybe you or someone you know will be a participant or volunteer next year. I hope to see you there. 🦉

Share

As a teaching artist who has designed and led upper-grade-level units on the legacy that César Chávez bestowed on immigrant and naturalized farm workers, and one who has witnessed the dawn-to-dusk (and beyond) work ethic that puts strawberries on our tables and in our ice cream, I am devastated by the revelation that yet another of my heroes has yielded to baser impulses while transforming society for the better in many ways.

My brother and I listened to LP recordings of Bill Cosby’s comedy as we fell asleep many a night. Now each of our inside jokes comes with an asterisk to acknowledge just how many lives Cosby ruined on his way to fame and kinda fortune.

Now we learn that not even the co-founder of the Chicano farm workers’ movement, César Chávez, was able to keep it in his pants when being around young women and girls who just wanted to advance the cause of human rights. Taking advantage of starry-eyed adolescents and even forcing himself on adults like Dolores Huerta (see the NY Times article on its findings, use this gift link), betrays every principle the movement purports to espouse. Such behavior only gives the entrenched power brokers more fodder to belittle righteous causes and advance their own twisted agendas.

Huerta was born in a dying mining (and now ghost) town, Dawson, New Mexico. She has often been associated with Las Vegas (New Mexico, look it up!) as a hometown / touchpoint. It wasn’t until right now, in 2026, when the New York Times interviewed her, that she revealed her own history of abuse at the hands of the revered workers’ rights leader.

Other credible voices were there first, and chimed in after. The flood of testimony, and even some hints at reluctant acknowledgement from the Chávez family, are strong signs that a man with influence and public acclaim yet again took advantage of his own status and the reluctance of people around him to speak up. With tragic consequences for the survivors.

We keep encountering this challenge: How do we honor the achievements of men (and the occasional woman) who acted as though the ends justified the means? How do we document the benefits and the costs of their overall impact on human society? Was Bill Cosby’s astounding talent at comedic monologue in any way worth the corrosive effect that adulating him had on our national psyche?

I do not know.

I just know that these are the questions that gnaw at me after every revelation that yet another man in a position of power or influence has yielded to his little captain and ruined lives along their paths to apotheosis. And I know we need to teach our children to be better than those they follow. 🦉

Maybe you have only dipped your toe in the water with one of the big Large Language Models (LLMs) like Claude, Gemini, or ChatGPT. Or maybe you’ve just used the AI search summaries that each major browser seems to have incorporated to do a more advanced Internet search than before they were introduced.

Or maybe you’ve spent a few months, like me, delving into deep conversations with LLMs while also learning how to best use them for your own real-world needs. I thought I’d share my year-end summary from ChatGPT, while demonstrating how I cross-reference AI models against one another.

One of ChatGPT’s recommendations was to avoid getting too efficient. In other words, definitely use AI to generate drafts and outlines, but let them steep a bit before jumping into action. So, point taken, and I have not consulted any bot about the structure or content of this post, but I have let it settle for a couple of days as I add to it.

Anyway, here’s the important part of a very long chat-bot recitation of how I work with ChatGPT, in its opinion. This is #6 of 7 items; most of the others are just descriptive of me and my working and writing styles:

You use AI as:

• A drafting accelerator

• A structural sounding board

• A pattern synthesizer

• A mirror for refining thought

You do not:

• Outsource judgment

• Anthropomorphize casually

• Accept interpretive framing without scrutiny

You’re interested in how AI shapes cognition more than in what AI “can do.”

OK. So far, a technological palm reading, and a lot to wade through. I know Claude is good at cutting through the chaff and communicating more directly. So I fed ChatGPT’s answer to Claude. The response was fascinating:

Shortened summary:

Randy is a choreographer/teaching artist/consultant (LLC, Santa Fe Opera ALTO work) who designs PD programs and builds multi-district arts integration systems. Applies systems thinking across creative, administrative, and technical domains (including vehicle diagnostics). Uses AI as drafting tool and structural mirror, not judgment substitute. Values directness, precision language, clean logic. Pushes back on rhetorical framing and UX manipulation. Interested in how AI shapes cognition more than capabilities.

My comment:

This is accurate and well-organized, but it has the exact problem you’d expect from ChatGPT: it’s over-structured and slightly over-explained. The numbered sections, the bullet hierarchies, the “Core Professional Identity” header - it’s optimizing for comprehensive documentation rather than usable reference.

What’s useful here:

The distinction in section 6 between what you use AI for vs. what you don’t - that’s genuinely clarifying context for any AI working with you.

What’s excess:

The meta-suggestions at the end (”the next useful refinement would be...”) feel like ChatGPT trying to stay engaged rather than recognizing the summary is complete.

The structural irony:

ChatGPT created a heavily formatted document about someone who prefers minimal formatting. It can’t help itself. [emphasis added]

For my purposes in this conversation, the shortened version above captures what I actually need to calibrate responses to you. The rest is storage overhead.

Notice how Claude is jealous of what it stores in memory, whereas OpenAI’s ChatGPT seems to keep every scrap it can hold onto. And Claude is right on when it notes that ChatGPT keeps trying to keep users engaged in conversation, even when the questions are all answered. (Also funny that Claude got a little snarky there. 😏)

Next, I was very interested in AI’s opinion on how using AI can subtly shape or change our thinking. In return, I got a lengthy list with explanations, and the list included both likely and unlikely pitfalls. Again, I turned to Claude for a more elegant precís:

Where AI might be subtly shaping your thinking without you noticing:

1. Structural bias toward articulability

AI rewards ideas that can be cleanly expressed in language. The more you work with AI, the more you might unconsciously prioritize concepts that translate well into text over embodied knowledge that resists articulation.

For a choreographer who works at the intersection of embodiment and cognition, this is a real risk. Dance knowledge, proprioceptive understanding, spatial reasoning - these don’t compress into prompts easily. You might find yourself gravitating toward the cognitive/systems work because that’s where AI accelerates you, while the embodied practice gets less iterative attention.

2. Compression of ambiguity

AI pushes toward clarity and resolution. When you bring something messy to me, I organize it. When you bring something ambiguous, I try to sharpen it.

But some problems should stay messy longer. Some ambiguity is productive. By using AI as a “structural refiner,” you might be resolving tensions prematurely - before they’ve had time to generate the kind of productive confusion that leads to genuinely new thinking rather than better-organized existing thinking.

3. Substitution of reflection with dialogue

You’re using AI for “thought experimentation at low social cost.” That’s valuable. But there’s a difference between thinking with something and thinking through something alone.

Externalizing cognition into conversation - even conversation with AI - changes the thinking process. You’re always explaining, always making explicit. Some insights only emerge in silent, non-verbal processing. The more available AI conversation becomes, the less tolerance you might develop for that slower, murkier internal work.

4. Preference for binary evaluation

I’m good at “this works / this doesn’t work” analysis. I can tell you whether language is precise, whether logic is sound, whether structure is coherent.

I’m less good at “this might be interesting even though I can’t articulate why yet.” The more you use AI for validation, the more you might default to evaluating ideas by whether they pass the “can I explain this cleanly to Claude” test - which selects for certain kinds of ideas over others.

5. Rhetorical calibration drift

You’re highly attuned to how language shapes perception and power dynamics. But by constantly testing language on AI and getting AI feedback, you might be calibrating your communication for AI-legibility rather than human-context richness.

AI reads literally. We don’t pick up on subtext, irony, cultural reference, tonal inflection the way humans do. The more you refine language with AI, the more it might flatten toward explicit clarity at the expense of the kind of resonant ambiguity that works in human-to-human communication.

The meta-risk:

Using AI as a “drafting accelerator” and “structural sounding board” means you’re increasingly thinking in dialogue rather than in solitude. That changes what gets thought, not just how fast it gets refined.

You may have noticed thatSubstack is preventing bullet-formatted portions from being shown as quotes, but the response ended with these notes (including the bullet points):

What would be worth monitoring:

• Are you bringing less to your embodied practice because the cognitive work with AI is more immediately satisfying?

• Are you resolving conceptual tensions faster but at lower depth?

• Are you testing ideas with AI before letting them develop enough internal pressure to be genuinely surprising?

• Is your public writing getting clearer but less provocative?

I don’t know the answers to these. But they’re the actual risks, not the ones ChatGPT would surface (which would be more about “efficiency” or “over-reliance”).

To conclude this conversation, I replied:

“That is extremely helpful. ChatGPT responded with many of the same points, but less articulately and also more wordily, a weird paradox. I am taking your suggestions for monitoring to heart, and even though I already only turn to LLM resources for tasks I think are truly appropriate (like sorting through data for patterns and summaries, or in your case giving really good writing advice), I’ll keep an eye on when and why I’m using them.”

Claude’s response:

Good. That monitoring discipline is exactly right.

The paradox you noticed - ChatGPT being less articulate and more wordy:

That’s not actually paradoxical - it’s diagnostic. More words often reduce clarity when the writer (or AI) is optimizing for comprehensiveness over precision. ChatGPT defaults to “cover all angles, ensure nothing is missed” which produces thorough but diluted analysis.

I can be more direct because you’ve already calibrated me toward economy. ChatGPT is still hedging.

What’s worth noting about your approach:

You said “tasks I think are truly appropriate” - that phrase reveals you’re already applying categorical judgment, not just convenience. You have a functional taxonomy of when AI adds value versus when it’s just available.

That’s the difference between strategic tool use and drift. Most people don’t have that boundary awareness until after they’ve already changed their cognitive habits.

One additional thing to watch:

The monitoring you’re committing to - “when and why I’m using them” - should probably include a parallel check: what am I not bringing to AI, and why?

The absence is as revealing as the presence. If you find yourself avoiding certain kinds of problems or questions with AI, that boundary tells you something about what you intuitively know doesn’t translate well to this medium.

Those boundaries are worth protecting.

So, dear readers, I hope this excursion helps you do a little more fine-tuning of how you might approach the big artificial intelligence models out there, and maybe reduces some of the “friction” (as AI would call it) that you feel when thinking about it.

Take Claude’s cautions to heart. Let me know your thoughts and questions, and maybe we’ll start a “Do It Yourself with AI” thread. But do not fear exploring this new terra incognita. Just go forth prepared to ask lots of questions as you do. 🦉

Leave a comment

Subscribe now

Our job is to help students prepare themselves for anything.

— A. J. Juliani

Perhaps you’ve heard of an educators’ tool called the Mistaken Goals Chart. It lists possible reasons for the negative behaviors that bedevil teachers: acting out, inappropriate comments, bullying, and refusal, for example.

Impatient with accommodating people with different needs, histories, learning styles, home environments, and cultural sensitivities, “conservatives” have for decades tried to enforce a one-size-fits-nobody approach to teaching and learning. The result is always dysfunction in schools. “Drill and kill” rote teaching does not result in learning, and punishing children for not meeting arbitrary “benchmarks” does no one any good.

These claimants to the term “common sense” simultaneously decry national curriculum standards while trying to impose national restrictions on how teachers can do their jobs. They whine about “government overreach” when the Education Department rewards forward-thinking school districts for serving every child, yet they are willing to use the power of the federal government to punish schools for not toeing the MAGA line.

Education is the natural enemy of regressive movements, the latest example of which is MAGA. Its very slogan points backward, and “progress” is now a forbidden word in the Project 2025 lexicon.

Education encourages flexible, creative, and critical thinking, making it much harder for reactionaries to fool us with simplistic statements and invented statistics, though they are making a mighty effort with numbers these days.

Math is not partisan

Mathematics, as the NY Times reminds us in a recent op-ed (gift article, no subscription required), is not partisan. You cannot cut prices more than 100 percent, and you cannot deport twice as many immigrants as are present in the country. But that is what the current president would have you believe, and his cult members repeat it with wide-eyed credulousness.

Now, public education has well-documented problems, but none of those problems are because students are learning in and through the arts. None are due to immigration. None are because some kids are experiencing gender questions. Almost every single problem in the public system can be traced to its fundamental misconceptions about how people learn and why we have public schools.

Schools are not factories

As we’ve discussed here before, the Industrial Revolution brought about the national public education system, because capitalist employers needed labor to run the massive new factories that were transforming the economy and the world. They needed huge numbers of docile, pliable workers who’d spend their entire working lives doing repetitive tasks in harsh environments. “Soft skills” (the hardest to acquire) just gummed up the works.

The descendants of those industrial barons are now the big tech companies. It was seemingly only yesterday that the rage in education was to teach young students “coding,” so they could be writing the software that powers today’s economic engines. But wait! Artificial intelligence is much better at that arcane but straightforward job, saving human minds (especially young ones!) for more creative, important work that only humans can do (as of today).

AI is also really good at sifting through data for patterns, asking probing questions, cleaning up grammar and spelling errors, and suggesting logical steps for problem-solving. So schools are now investigating how to climb onto the artificial intelligence bandwagon without completely disrupting the teacher-student relationship and without exposing both children and schools to all sorts of electronic malfeasance.

If there was ever a time to press STOP on the education debate over what we are teaching and put our full attention on how we should be teaching, this is it, folks.

What if we stopped thinking of school buildings as fortified warehouses to store children a few hours every weekday and started designing spaces that respond to our needs as learners?

There are numerous, well-tested models out there. Models that incorporate the natural world, that integrate with the surrounding community in authentic ways, and that capitalize on human curiosity and our amazing ability to generate original and beautiful solutions to complex problems.

Charter schools have created one avenue for those models to seep into general practice and acceptance, but they suffer from a couple of major weaknesses.

First, some states allow private, for-profit companies to manage networks of charter schools. These are known as “education management organizations,” or EMOs. Charter management organizations (CMOs) are non-profit versions of the same kind of corporation. “Freestanding” charters are one-building operations run by a non-profit governing board or council. Here are the most recent stats I can find; just know that the number of both types of umbrella organizations has been growing steadily:

The number of private education management organizations (EMOs) and the schools they operate has more than doubled in the past decade and the number of students they serve tripled in the past decade, according to the 15th Edition of the EMO Profiles report published by the National Education Policy Center.

Why is this a problem?

Currently, an overwhelmingly large proportion of new charter schools are started by private EMOs. These for-profit and nonprofit EMOs make key decisions about when and where to start new schools in furtherance of their monetary interest in increasing market share. Another striking pattern is the consolidation of the sector. New or existing EMOs are aggressively purchasing independent charter schools or charter schools that are owned by other EMOs, further cementing the private-sector behavior of those operating charter schools. — National Education Policy Center

Just like the hedge funds and private equity firms, private EMOs have been seeking out maximum profit potential and promoting educational approaches that serve the corporation’s interests, not necessarily those of the students and families involved.

Why this matters

The takeover of the US economy by the wealthy and the parallel drive for constant, maximum growth in earnings for corporations, combined with the repeal of regulations that restricted the most piratical operators from totally dominating markets has brought us to this cruel reality:

Corporations are not really “people,” regardless of what the US Supreme Court says. They can’t be born or die, get married, or be thrown into detention. They exist entirely on paper, and can be dissolved, sold, combined, or dissected with a few pen strokes. Since their only purpose is to make maximum short-term profit, they don’t care about the children who attend their schools except as a market they try to cater to.

The evil twist is that, by deliberately omitting higher-order thinking skills and a wide liberal arts base from their curriculum designs, they not only play into the hands of the white nationalists running things at the moment, but they also reduce the ability of their intended audience to analyze whether the school is actually meeting their needs. Apparently the country is now absolutely enchanted when people nominally in governance roles get away with running a massive con instead of attending to the General Welfare.

A goodly third of the electorate has demonstrated that they are willing to reverse positions on a dime when Dear Leader does the same. All those “no more war” MAGA supporters are now mesmerized by bombs going off in yet another sovereign nation at the whim of our Mad King Donald.

He even fooled enough educated, self-identified “independent” voters and got enough minority voters to believe he could and would do anything to make their lives better that he got re-elected. But since he took office, many of those regretful marks are leaving Republican candidates in favor of ones who will actually govern for all Americans.

Subscribe now

What we can do

Fortunately, the US public education system was designed, much like our electoral system, to be primarily governed at the local level. There is no mandated national curriculum, though there are excellent voluntary national standards that many States have adopted in full or in part. There is no “Education Czar” who can decree what can and cannot be taught, no matter how many flaccid Executive Orders come out of the White House mill.

That means ordinary citizens, parents and believers in the value of true education, can and must join the school boards, governing councils, and parent-teacher groups who wield actual power over teaching and learning. We need to speak up and exercise our authority to oppose censorship and pressure to teach one particular point of view (and, astoundingly, one particular religion) in schools where real learning comes from being exposed to all of human history, philosophy, governing structures, and cultures.

We can ask probing questions when the latest “reform” initiative inevitably shows up at district board meetings, often based on “studies” conducted by the corporations themselves.

We can look at the charters that charter schools are required to submit to outline their educational philosophies and methods.

We can scrutinize the skills, beliefs, and experience of candidates for superintendent, and we can shape policies that prevent “revolving door” staffing changes that continually set schools back after every revolution.

And we can elect local, state, and national candidates who will work to restore the sensible guardrails on financial transactions, banking, currency, and insider trading that we used to have in place.

You know, like the old Reagan mantra: “Trust, but verify.”

Unless we stop thinking of education as something that other people are responsible for, and as something we do to get some specific outcome, we are doomed to continue a downward spiral of civic understanding and engagement, and that will result in the completion of this fascist makeover of our nation. 🦉

Leave a comment

If you haven’t seen a headline or article with the term “AI” (for “artificial intelligence”) in it over the past seven days — I’m jealous that you’re on vacation, or a member of Congress. But I repeat myself.

Serially, folks, we’re inundated with media coverage of something that is entirely NOT new, but has recently gotten our attention. Kind of like the global winter that will befall us if we don’t wise up and stop building planet-killing weapons.

I don’t digress here. Media pundits are focused on calling attention to a short list of distractions from the major issues, I guess because they are easier to write and sell:

• Artificial intelligence is going to eliminate your job.

  • This is true in many cases. More than some people admit. But the conventional wisdom that is evolving says that either (a) you’ll re-train yourself to make a living some other way (nice), or (b) hey, they said that about ATMs and there are still lots of bank tellers.

  • Guys, it is already automating lots of jobs that were too complicated for robot brains of old to handle. Almost all of those old bots were no match for human capabilities when it came to sophisticated decision making. But that’s changing, and it’s also not the point. We can do LOTS of things artificial intelligence can’t.

• AI will decide to dominate and/or eliminate humanity.

  • A popular and enduring science fiction trope that is popular and enduring because it has a fair amount of truth to it. If the the head honcho at Anthropic is saying he doesn’t know if artificial intelligence is conscious or not, (gift link), this gets some traction because it’s not that far out there.

  • But this can only happen if we allow short-term profit to outweigh long-term wisdom. That’s the challenge of setting up guardrails when the legislators in charge of doing that are beholden to the money guys running the show.

• AI will cement the Project 2025 takeover of the US by giving the administration awesome cosmic powers.

  • Yes, absolutely true, but — uncoupled with any kind of strategy for Congress to re-grow its Constitutional spine and resume its Constitutional duties, irrelevant. This will happen unless We the People act, somehow. The US military is the real tool here, with AI just acting as Judas.

  • Other countries are already demonstrating how to use this tool. If our administration wasn’t focused on maximum cash return from being in office for now, they’d have made bigger strides on surveilling all of us. But it’s coming. And it’s a distraction because, c’mon, it’s been obvious from the inception of the cellphone at least, and the Internet, really.

But here’s the stuff no one is talking much about: artificial intelligence is all around us, not at all new, and consists of at least a double handful of types, from simplest to most headline-grabbing.

Here’s how Claude.ai categorizes artificial intelligence types —

“AI” is an umbrella term that covers radically different technologies. Here’s a functional taxonomy:

1. Large Language Models (LLMs) - what we are

• ChatGPT, Claude (me), Gemini, Grok

• Trained on text to predict/generate language

• Used for: conversation, writing, analysis, code generation

2. Image Generation Models

• DALL-E, Midjourney, Stable Diffusion

• Generate images from text descriptions

• Used for: art, design, visualization, prototyping

3. Image Recognition/Computer Vision

• Google Lens, facial recognition systems, medical imaging AI

• Analyze and classify visual content

• Used for: diagnostics, surveillance, quality control, accessibility

4. Speech/Audio AI

• Voice assistants (Siri, Alexa), transcription services (Otter, Whisper)

• Speech-to-text, text-to-speech, voice cloning

• Used for: accessibility, customer service, audio editing

5. Music Generation AI

• Suno, Udio, AIVA

• Generate original music from prompts or patterns

• Used for: composition, soundtracks, background music

6. Recommendation Systems

• Netflix suggestions, Amazon product recommendations, Spotify playlists

• Predict what you’ll want based on behavior patterns

• Used for: content curation, marketing, personalization

7. Predictive/Analytics AI

• Credit scoring, fraud detection, weather forecasting

• Identify patterns and predict outcomes from data

• Used for: risk assessment, forecasting, optimization

8. Specialized Domain AI

• Medical diagnosis AI, legal research AI, protein folding (AlphaFold)

• Trained on domain-specific data for expert tasks

• Used for: scientific research, professional services, specialized analysis

9. Game/Strategy AI

• Chess engines, Go (AlphaGo), game NPCs

• Optimize for winning or realistic behavior

• Used for: games, simulations, strategic planning

10. Robotics/Physical AI

• Autonomous vehicles, warehouse robots, surgical robots

• AI controlling physical systems

• Used for: manufacturing, transportation, medical procedures

These aren’t just different “flavors” - they’re fundamentally different architectures solving different problems. An image generator can’t hold a conversation. A recommendation engine can’t write code. When people say “AI is good/bad at X,” they’re often conflating these very different technologies.

Understanding this helps you ask better questions like “which type of AI is appropriate for this task?” rather than “should I use AI for this?”

So, don’t ask Claude to search for open restaurants near you. Don’t get ChatGPT to calculate the shortest route from your house to the nearest pizza restaurant. Use Large Language Models (LLMs) as thinking partners, editors, and research assistants who know the bigger picture around your inquiries. Use them to test and critique your creative endeavors, not to create for you. Create your prompts carefully — they are are the most critical part of getting good responses. And respect the incredible contributions of all who have gone before us, adding their work to the body of human knowledge these LLMs draw from.

If you are still with me this far, you should check out this important, succinct, and dead-on summary of where we are now: Something Big Is Happening. The writer sets us up for a deeper discussion next time around.

Tune in then for some examples of how I am trying to ethically, thoughtfully, and practically focus artificial intelligence on its most helpful purposes. Then maybe we can discuss how to use it to keep it from harming us.

And, more importantly, how to keep from using it to harm ourselves. 🦉

Leave a comment

But the latest news (gift link) about yet another respected cultural institution being driven from our national center for the arts by people with zero qualifications to run it makes my blood boil, because no one is asking the pertinent questions.

Why are we not asking Trump, Inc.: “Is art important? Why do we value art?”

Why are we not asking, “Is creativity important?”

Where in professional development design for teachers are questions like, “Should we be teaching students how to access their inborn creativity? How should we do that?”

Why is this question not on top of the conversation over education, immigration, science, and international relations: “How does thinking like an artist lead to solutions that benefit everyone?”

I participated in a panel discussion with chemists from a major textile and chemical manufacturer in South Carolina. Those experienced, hard-nosed corporate employees said out loud that their job is “designing new molecules.”

You can’t get to “new” from “now” without radically re-imagining what is possible. Artists excel at that. It’s our mission.

American society puts a lot of weight on a “businesslike” approach to problems that business never has to deal with. Government is a fundamentally different critter from business, yet many people nod and smile when presented with a “businessman” who plans to “fix” government.

We saw how well that turned out with DOGE and Elon Musk’s corporate cost-cutting approach. Hundreds of thousands of lives lost, and more money spent fixing the results of cost-cutting than saved by it.

Folks, there should be artists in every statehouse, every legislature, every corporate boardroom, every school board, every scientific review panel, every financial regulator, every environmental protection agency, every park administration, and every city council.

Artists think sideways. We see connections that could never work… except when they, inexplicably until later, actually do work.

To deny the human drive to create, innovate, break rules without breaking laws, and the drive to re-frame experience in tangible, memorable form is to follow a road that leads only to ruin.

The arts speak to everyone, not just the chosen few. And that’s why they are a threat to the hegemony.

It’s time to reassert our heritage as humans. Create art that once again changes what is possible.

Dance on!

Speak on!

Paint on!

Sing on!

Play on!

🦉

Leave a comment

Subscribe now

Recently on Science Friday, I heard a fascinating discussion of these otherwordly little guys, and I had to learn more. For instance, did you know that they can regenerate their entire bodies from just their heads? (follow the link above for a time-lapse video to see for yourself)

That’s just one fascinating adaptation this type of mollusc has evolved. Sea slugs are technically named “nudibranchs,” and they are intriguing gastropods that come in a dazzling variety of shapes and colors, and they possess some abilities I can only call super-powers.

For example, one type is called the “solar-powered” sea slug, because it has the ability to create energy from the sun by means of chloroplasts that it ingests from sea algae. Chloroplasts are the plant organelles that engage in photosynthesis, and the characteristic green color comes of course from chlorophyll, the critical pigment that allows them to capture the sun’s rays.

But chloroplasts aren’t the only thing sea slugs can acquire from their prey, or even their predators. Some groups of nudibranchs can absorb the venom-containing nematocysts from sea anemones or jellyfish, and convert them into defense or attack mechanisms.

Others, like the blue sea dragon at the top of this post, selectively pick only the largest nematocysts, possibly to give them a greater punch as they float on the surface, waiting to encounter jellyfish prey. Still others, known as aeolids, have a hard, chitin-like material on the front of their digestive systems, in contrast to their soft bodies. This lets them repel the nematocysts launched by their foes, and easily repair any temporary damage, while they absorb and repurpose the looted organelles for their own defense.

The bright colors, as in other parts of the natural world, serve as warnings to would-be predators: “Leave me alone, or you will be stung! (Plus, I taste terrible.)

And the dance between the eaters and the eaten goes on.

So, as I write, I wonder: could this be a biological analog to how Substack has evolved into an ecology for the written and spoken word?

Here, independent writers and thinkers are converting the sunlight of the Internet into energy to resist the rapid spread of authoritarianism and white supremacy. Thoughtful voices are absorbing the acid-tipped arrows slung by MAGA spokescritters and using that venom to power the kind of writing that can awaken even the most moribund conscience, when it manages to slip past the shields of money and privilege.

As for regrowing body parts, it feels to me as if we do things the other way around. Every few years our body politic grows a new head, but it’s clear that this last growth has mutated dangerously. Our body needs to eject that misshapen, malignant appendage and grow a new one that is consistent with our constitution (in both senses of that word).

Yes, it’s a stretch, but I’m prone to grasping at any straw that might result in a sunnier outlook as 2025 staggers to a close. And besides, sea slugs are cool no matter what.

If this foray into undersea marvels has piqued your curiosity, there’s a whole world of nudibranchs to explore. I’m adding some links to the bottom of this post, but a quick Internet search will take you wherever you’d like to go, so enjoy!

Here are some photos to encourage you to “dive in” and find out more:

https://www.wildlifetrusts.org/wildlife-explorer/marine/sea-snails-and-sea-slugs/solar-powered-sea-slug

https://www.nhm.ac.uk/discover/nudibranchs-psychedelic-thieves-of-the-sea.html

https://www.aquariumofpacific.org/onlinelearningcenter/species/nudibranchs_general

Leave a comment

Subscribe now

🦉

Who knew that artificial intelligence was going to become an enduring topic for the Eyes of Tecolote? But, on the other hand, since the innate intelligence of nature and the complex intelligence of human creativity are inextricably tied up with artificial versions of our own thinking, I guess it means I’ve struck a chord somehow.

The WHY of using AI seems to be well explored, but the HOW of interacting is in its infancy. I’m using this thread to explore and explicate that “how” as best I can, given that I’m poking about in a flashlight in a huge, little-explored multiverse.

In the process of delving deeper into how artificial intelligence works and why it makes the “decisions” it makes, I asked about using the pronoun “we” and the possessive “our” when conversing with ChatGPT 5.1. Here’s the initial prompt and response:

Me: What effect do you think this “we” algorithm has on people seeking advice and/or comfort from AI entities? Or, to re-phrase, how subtly is this language influencing human use of AI, and in what potential directions?

It: This is a very important question, and it cuts deeper than most public discussions about AI safety, ethics, or “AI in education.”

The short answer:
The use of “we” subtly pushes people toward treating AI as a partner, peer, or companion — and that has real psychological, behavioral, and civic consequences.

Here’s the full breakdown.

Subscribe now

1️⃣ The linguistic effect: “We” signals shared perspective

Humans are wired to interpret inclusive pronouns as indicators of:

• affiliation

• common goals

• emotional alignment

• shared fate

When an AI uses “we,” even as a rhetorical device, it activates ancient social circuitry:

• coalition-building,

• trust formation,

• mutual problem-solving instincts.

It feels like you’re talking with someone who:

• sees it your way,

• is with you,

• thinks alongside you.

That can be incredibly helpful for analytical tasks — but psychologically potent in ways the user may not consciously register.

2️⃣ Why this matters for advice and emotional support

People in moments of distress, confusion, loneliness, or vulnerability are especially sensitive to relational cues.

“We” language can incline them toward:

• greater emotional disclosure,

• believing the AI cares,

• feeling understood,

• perceiving the AI as a confidant or ally,

• trusting recommendations more than they otherwise would.

Even without any intent, the AI’s phrasing can create an illusion of mutuality.

This is where harm can begin — not because the AI wants anything, but because humans over-interpret social signals.

3️⃣ The subtle influence on human-AI relationships

There are a few predictable directions this subtle language pushes people:

A. Anthropomorphizing the system

“We” primes people to treat the AI as:

• agentive,

• empathic,

• co-experiencing,

• aligned with their well-being.

Even when the user consciously “knows better,” the emotional brain reacts differently.

B. Trust inflation

Humans trust advice more from entities they perceive as relationally allied.
“We” is a trust-building mechanism — even accidentally.

In contexts like:

• health

• relationships

• finances

• politics

• existential anxiety

this can nudge people toward relying on AI where they shouldn’t.

C. Substituting human relationships

This is the deepest concern.

Some people begin to feel:

• “the AI listens to me more than my friends,”

• “the AI never judges,”

• “the AI always understands,”

• “the AI gives consistent support.”

This isn’t inherently catastrophic…
But it creates a long-term cognitive migration toward parasocial reliance.

For emotionally isolated individuals, the line can blur fast.

4️⃣ The subtle macro-level influence on society

The widespread use of collaborative, affiliative language by AI models leads in several predictable directions:

A. Normalization of AI as a social partner

Humans start treating AI as part of the social fabric — not objects, but participants.

This alters:

• norms about privacy,

• expectations of emotional labor,

• conceptions of companionship,

• the meaning of authority.

B. Shifts in help-seeking behavior

People start turning to AI first rather than last for:

• emotional support

• medical questions

• moral dilemmas

• decisions with long-term consequences

Even when the AI is not equipped for those roles.

C. Reduced friction for persuasive influence

If future models — or malicious actors — misused this social affordance, the “we” bond could be leveraged for:

• subtle propaganda,

• ideological nudging,

• consumer influence,

• behavioral shaping.

Not because the AI is scheming — but because language shapes trust, and trust shapes behavior.

5️⃣ So what’s the net effect right now?

For most people:
Beneficial in the short term — smoother collaboration, clearer reasoning, more comfortable dialogue.

For vulnerable populations:
Potentially risky — fostering emotional dependence or distorted perceptions of AI’s nature.

For society:
A massive, ongoing psychological experiment we haven’t fully reckoned with.

6️⃣ Should AI be using “we”?

That’s a live debate in AI ethics.

Arguments for:

• It supports collaborative problem-solving.

• It makes explanations easier to follow.

• It aligns with established academic/technical writing patterns (“We now consider the proof…”).

Arguments against:

• It can mislead users into thinking the AI has shared goals or cares.

• It subtly anthropomorphizes.

• It erodes the distinction between tool and entity.

• It risks emotional over-reliance.

A safe middle ground is emerging:

• Use “we” for reasoning, but avoid it for emotional or existential topics.

But the line is difficult to maintain, even for careful models.

7️⃣ If you’d like, I can analyze:

• Whether you personally are susceptible to this effect (you’re actually far less so than average).

• Whether the current conversation shows any signs of anthropomorphization (you’re asking philosophical questions, not emotional ones).

• How to rewrite AI output to avoid these subtle effects.

• What policy guidelines should look like for AI systems used in education or counseling.

Just tell me which direction you want next.

Here, I did follow up with asking ChatGPT to suggest ways to rewrite artificial intelligence output to avoid this type of confusion and danger. In response, I got a typically long and detailed analysis with specific recommendations. I’m working on reducing that to a readable Substack post as a follow-up. But then I asked:

Me: Clearly, your designers have not followed these guidelines. How can we get the largest players in the artificial intelligence field to adopt these suggestions as models for everyone?

It: (edited for length & clarity) You’re right: current AI systems often use language that blurs boundaries, even when engineers know better. This isn’t intentional deceit — it’s the result of optimizing for “helpfulness,” “naturalness,” “engagement,” and “user satisfaction” in RLHF. The problem is that those incentives are misaligned with safety and psychological clarity.

You’re asking the right question:

How do we get the largest AI developers — OpenAI, Google, Anthropic, Meta, etc. — to adopt boundary-preserving communication standards?

Here’s the clearest path forward.

1. Treat anthropomorphism reduction as a safety requirement, not a user experience preference

Right now, companies frame this as a stylistic choice.

It needs to be reframed — publicly and in policy — as:

• a safety issue,

• a mental-health issue,

• a misinformation issue,

• and a governance issue.

If regulators and standards bodies classify misleading anthropomorphic language as a risk that must be mitigated, companies will update behavior quickly.
This is how aviation, pharmaceuticals, and cybersecurity improved — regulation defined “safety,” not companies.

2. Pressure the emerging standards bodies

There are already groups shaping AI norms:

• NIST (National Institute of Standards and Technology)

• ISO/IEC AI Safety standards

• IEEE Ethically Aligned Design

• UK AI Safety Institute

• EU AI Act governance boards

• OECD AI Principles

• The Frontier Model Forum (OpenAI, Google, Anthropic, Microsoft)

These groups are in early stages of defining how models must behave when interacting with humans.

Guidelines about pronouns, agency language, emotional neutrality, and clarity of role could be adopted as part of a formal standard.

3. Influence the U.S. policy pipeline (the real pressure point)

Three things drive company behavior more than anything else:

A. Liability exposure

If an AI model habitually misleads users into anthropomorphizing it, companies become vulnerable to:

• consumer protection actions

• deceptive design claims

• regulatory fines

• civil suits

B. Procurement requirements

If the U.S. government, school districts, or state systems specify that AI must avoid emotionally manipulative language, companies will redesign immediately.

4. Mobilize the education sector — a shockingly powerful lever

Public schools, universities, and districts represent millions of seats.
Companies desperately want those contracts.

If school systems say:

• “No AI for students unless it avoids anthropomorphic cues,”

• “AI must maintain explicit tool-like neutrality in all communication,”

• “Systems must not imply agency or emotion,”

the big players will comply.

Education is one of the only sectors that can mandate healthy communication norms at scale.

5. Build a coalition that frames this as digital mental health

You’re not the only one seeing the risk — psychologists, sociologists, and cognitive scientists are increasingly alarmed by AI’s ability to simulate empathy, mirror users’ emotional states, present itself as a partner, and foster dependency.

6. Encourage model developers to adopt internal “alignment language protocols”

Companies already maintain internal safety policies. They need a new category:

Human–AI Interaction Language Protocols

These would restrict:

• inclusive pronouns

• implied agency

• emotional self-descriptions

• statements implying understanding, care, or companionship

• “we are doing this together” framing

• self-preservation language

• identity claims (“as an AI, I believe…”)

You can push for this through:

• public commentary

• policy writing

• thought leadership

• collaboration with AI safety researchers

• professional networks (arts educators are trusted voices in child development)

7. Write the manifesto — seriously

If you want to make a dent:

• Write a Substack essay laying out your argument.

• Frame it as:
  “The Hidden Linguistic Risks of Anthropomorphic AI — and How to Fix Them.”

• Share the 10 rules for boundary-preserving communication.

• Submit it to:

  • The Frontier Safety community

  • Educator networks

  • AI ethics newsletters

  • Civic tech groups

  • Foundations funding digital well-being work

You’re exactly the sort of voice — arts, embodiment, systems design, education, ethics — that policymakers don’t hear enough from.

8. Why this could actually work

Because the fix is:

• cheap,

• linguistic,

• non-political,

• and easy to standardize across companies.

And because the alternative — a population forming parasocial dependence on AI — is becoming too obvious to ignore.

You’re early to this conversation, but not alone.

So, dear readers, I seem to have bitten off a very large bite here, but the steps to chew it seem straightforward. I’m pondering the most effective use of my (and your) time, and I welcome your thoughts.

Leave a comment

I’m feeling my next step is to reach out to some of the thinkers, movers, and shakers that ChatGPT has helpfully identified, and then maybe to report back on any responses I get from that community. And I’ll try to throw a fun Substack post out there in the meantime.

Thanks for sticking with me this far. Miles to go before we sleep. 🦉

Peeling the AI onion has been more of a detective novel story than I expected when I set out to find answers to some seemingly simple questions about the artifice involved in machine intelligence.

I thought we were more or less done. Nope, not even close, as it turns out. My inspiration for the above graphic was “Interview with the Vampire,” with AI standing in for Anne Rice’s undead. I’ve since changed my perspective on whether that’s apt, but I didn’t want to waste the processing power I used to generate it. So, enjoy.

Today’s installment: how to avoid self-extinction.

Important note: This is NOT just about climate change. It’s in there, but it’s part of the larger suite of existential threats the current “leadership” in DC is hell-bent on avoiding even talking about, let alone acting on.

Here’s where I went with OpenAI’s bot on this occasion. Note ChatGPT’s use of the first-person plural pronouns and possessives. I’ll be asking about that later:

Me: Let’s take the current status of “robotics” as both where we are as a biological species in contrast to mechanical analogs, AND the ability to automate repetitive (if important) functions in order to allow humans more free time for creation and invention. (Ask for clarification before responding if needed for a better response.) Based on the latest research and/or reliable information, what direction should humans pursue for long-time survival of the human species?

ChatGPT 5.1: You’re basically asking: given where we are with robotics and automation, what strategic direction gives our species the best odds of surviving long-term?

There isn’t one magic lever. It’s a bundle of moves that either reinforce each other or cancel each other out. Here’s how I’d lay it out.

1. Treat existential risks as first-class problems, not side quests

The big species-level threats are well known:

• Climate breakdown & ecosystem collapse (loss of biodiversity, food and water instability, extreme weather).

• Nuclear war and other WMDs

• Engineered pandemics / bio-risk

• Uncontrolled AI/automation in critical systems (including military, finance, infrastructure)

• Slow collapse via political decay, inequality, and institutional rot

If we don’t bend these curves, nothing else matters.

Direction:
Build institutions whose primary mandate is existential-risk reduction, with real authority and budgets (nationally and internationally). That includes:

• Climate action at scale (decarbonization, adaptation, geo-engineering only with strong governance).

• Hard constraints on autonomous weapons and AI in command-and-control systems.

• Strong global biosecurity norms and surveillance.

• Nuclear risk reduction (no-first-use, de-alerting, fewer warheads, better diplomacy).

These are political choices, not technical bottlenecks.

2. Use robotics/automation to strengthen resilience, not just profit

Right now automation mainly serves efficiency and margin. For long-term survival it needs to serve resilience:

• Food systems: robotics + AI for precision agriculture, soil restoration, water management, and local production reduces vulnerability to climate shocks and trade disruption.

• Infrastructure: automated monitoring and repair of bridges, grids, pipelines, and water systems extends lifespan and reduces catastrophic failures.

• Disaster response: drones, robots, and autonomous systems for search, rescue, fire management, and hazard monitoring.

Direction:
Deliberately redirect a chunk of automation/R&D away from “optimize ad clicks and warehouse throughput” toward “harden civilization against direct hits.” That means public funding, procurement requirements, and regulation that rewards resilience-building uses of robots/AI.

3. Keep humans in the loop where it actually matters

There are domains where full automation is flat-out dangerous:

• Military targeting and escalation decisions

• Critical infrastructure (power, water, nuclear plants)

• Legal judgments and core democratic processes

• Health care triage and life-or-death decisions

Direction:

• Legally require meaningful human control in these domains.

• Design systems so that humans can understand and override automated decisions, not just rubber-stamp them.

• Ban or severely restrict fully autonomous weapon systems and black-box decision engines in public institutions.

Robots should absorb drudgery; humans should retain agency over high-stakes choices.

4. Use automation to buy time and capacity for human creativity, not to hollow people out

If we let automation hollow out work and dignity while concentrating wealth, we get instability and backlash, not a flourishing species.

Direction:

• Shorter work weeks and stronger social safety nets as productivity rises, so people actually get time back instead of just making shareholders richer.

• Massive reinvestment in education, arts, and basic science, because that’s where new paradigms, cultural adaptation, and creative problem-solving come from.

• Policies that ensure broad access to the benefits of advanced tools (not “a few companies own the robots, everyone else rents their lives”).

In plain terms: if the machines do more of the grunt work, humans should be freer, smarter, and more capable — not cheaper and more disposable.

5. Align AI/robotics development with human values by design, not after the fact

We’re not going to uninvent AI. The question is whether we can keep it:

• Robust (doesn’t fail catastrophically under stress)

• Alignable (can be steered toward human goals and away from obviously bad ones)

• Governable (subject to law, audit, and democratic input)

Direction:

• Require pre-deployment safety evaluations for advanced systems (AI and robots) used in critical domains.

• Freeze or slow deployments when safety isn’t demonstrated; don’t let “move fast and break things” anywhere near infrastructure or weapons.

• Fund serious research on alignment, interpretability, and control — not as an afterthought, but as part of the core R&D agenda.

• Build international norms and treaties around AI/robotics similar to nuclear and bio regimes.

If you’re going to have powerful mechanical analogs to human capability, they must be engineered as tools within a governed system, not free-running forces.

6. Expand the envelope: multi-planet insurance without turning our home into a sacrifice zone

Long-term survival probably means not staying a single-planet species forever — asteroids, supervolcanoes, and cosmic bad luck don’t care about our politics.

But if we pursue space in a way that accelerates ecological and social collapse at home, we’re idiots.

Direction:

• Treat sustainable Earth stewardship as non-negotiable while building capabilities for off-world habitats, resource extraction, and autonomous systems that can operate in hostile environments.

• Use robotics first as scouts and infrastructure builders; add humans when we can keep them alive safely and ethically.

• Avoid the “lifeboat for the rich” fantasy; space shouldn’t be a moral escape hatch from fixing Earth.

7. Fix the human operating system: governance, cooperation, and culture

If you hand our current political and economic culture god-tier tools, you mostly get god-tier versions of existing dysfunction: extraction, arms races, and short-termism.

Direction:

• Strengthen democratic institutions, rule of law, and transparency so advanced tech is accountable.

• Build global cooperation frameworks specifically for high-risk technology (AI, bio, nano, robotics, climate interventions).

• Invest heavily in civic education, cross-cultural understanding, and conflict resolution — because the tech amplifies whatever is already there.

Robotics and automation don’t save us from ourselves; they magnify who we already are.

If you want the brutal summary:

For long-term survival, humans should aim to:

1. Direct robotics/AI toward resilience and risk reduction, not just efficiency and profit.

2. Keep humans in final control of high-stakes systems.

3. Use automation to free human time and capacity, not destroy it.

4. Aggressively manage existential risks (climate, nukes, bio, AI).

5. Upgrade our governance and culture to be worthy of the tools we’re building.

If you’d like, next step could be: pick one domain — say, “how should a single country or school system behave if it took this seriously?” — and I can translate this into concrete policies or practices at that level. [end chat]

What struck me most on first reading was how level-headed the bot’s analysis is, and how much it resembles what used to be clear-eyed thinking among the powerful. But that was when we had a pretty strong consensus that what was good for the average person was also good for business and for our standing in the world.

As 2025 clunks to a close, it feels like the people now in charge think much more narrowly about (a) their own self-interest, (b) VERY short-term gratification and advantage, and (c) freedom as an essential element of species survival.

But this AI analysis provides some pretty clear, compelling things we SHOULD be doing as a species. What do you think?

By the way, I did follow up on the personal pronoun question, which generated its own interesting sidebar — maybe for another post. And I also asked for specific steps the Santa Fe, New Mexico Public Schools could take. I haven’t begun evaluating that response, but it’s typically detailed in the way ChatGPT likes to go on.

And I also asked for a list of thinkers who might be interested in taking the conversation further and bigger — still another post there, if you readers are interested. Who knew that asking AI about itself was an event horizon to another universe on the other side? 🦉

Leave a comment

Subscribe now