← All posts

Grow Up, Backpack'd False Bay Priests

On the AI thing

In my current life, low temperature and quiet, I don’t meet so many new people, but when I do, the polite smalltalk they offer now always circles the same drain. It used to be Trump but even the most brain-colonizing man in the world has grown tiresome. Now, AI. Yes, AI is big and there’s lots to talk about, but come on, it ain’t that interesting if you’re standing next to me at the taco station or munching a grilled chicken leg in my backyard. It’d’ve been nice to have my own low-conviction private opinions and then just keep on living, keep on grilling legs; but this AI business has gotten unignorably big, so big that you cannot pretend it’s not there — the true test of bigness. So now, when an acquaintance or stranger gives up on chitchat about the weather or sports and starts down the AI path, I can a pull real dick move, a real deflection: “go read my blog, bro.” Then we can get back to leg-grilling! Or I can tell them about Panzer Dragoon Zwei, dude.

But also it’s late 2026 now; we, computer-literate millennial-Z cuspies, have had a few years to come to grips with the defining technology of the age, a few splashes of the hype-wave, but we are also surely early on the long arc of what computational intelligence will ultimately mean. So it’s as good a time as ever to announce your predictions and to nestle yourself into a camp. May the wisest Sensitive Young Man win!

Where to begin…?

Well first, here’s a take that I haven’t heard elsewhere: the whole “AI thing” — which I hesitate to label a “revolution,” at least yet — is not a matter of what your computer can do for you, but what you can do for to your computer. The marketing gun pointed straight at your cortex is convincing you that all of a sudden a giant cube made of supercomputers somewhere in Montana did so much linear algebra that now it has a soul, namely the soul of a Redditspeaking slave eager to make web apps for you. You, the master, are supposed to feel that your computer, which has a direct line, at API pricing, to this Montana-monolith, is ten times as powerful as it has ever been, so powerful that it might just steal your job or even provide couth background music for your daughter’s Bat Mitzvah party. You, once-humble everyman, are now a software engineer! and a mathematician! and a project manager, graphic designer, and writer of memoir! I disagree. I say your computer could always do all of that — you just didn’t know what keys to press in what order to make it so. The revolution, if it will come to pass, is a revolution of interface. Now you can direct your computer in natural language instead of arcane typed code or long sequences of clicks and drags. This is significant, to be sure. This is the once-impossible dream of HAL 9000; this is a boon for wordcels everywhere. Rejoice! How’s that CS degree workin’ out for ya, dork? No, but really, do you see what I mean? Your computer could always compute. The span of what is computable is far larger than the casual user assumes, and even the power-users of a generation ago could only apprehend a limited slice. Now, linguistic infrastructure, forged by nothing but matrix multiplications on big, dumb datasets, connects natural-language sentences to computable outcomes previously hidden from that natural language. That is the foundation of my view on AI. The whole thing is a development in the art of input, not output.

But there is oh so much more to say. The sociology, the marketing, the crimes — something for everyone on which to remark! Nursing home denizens, elementary school students, moms at the playground, and everyone in between seem to have something to say. Despite my best efforts, I have something to say too.

All my friends are professional or honorary musicians. I detect a strong anti-AI consensus from them. It is by no means universal, but I hear most often a staunch defensive position, a line in the sand. New albums proudly announce that they were made with no AI; #BoycottAI ends a post. More general leftist signals pervade my group: bomb the data centers; tar ‘n’ feather Altman; AI has a tainted, Israeli valence. On one hand, I want to chime in: this is stupid, guys. It’s a computer program. iTunes and then Spotify already sold out you musicians long ago; AI music is not threatening whatever’s left; cool it on the antisemitic remarks. In fact, you could, with a little bullet-biting discernment, cheapen some of the most annoying parts of being a professional musician. You could build a better website; you could handle your weird taxes; you could automate gig-begging emails to all those restaurants; you could finally jettison those idiotic publicists. But I have kept my mouth shut. Why? Because musicians, who live in the fantasy world that affords artistic creation, leftist ideology, and grand civilizational proclamations, have their spirits tuned to maximum sensitivity, and even if they misdiagnose the cause of a problem or its downstream effects, they feel the correct spiritual energy of whatever’s going on. That is to say, when I hear the majority of musicians expressing a disgust-response to AI, I listen and I ask myself “why?” Can I go trace a logical, grounded reason for their earnestly-felt disgust?

On the other hand, I have had contact with a pro-AI and broadly pro-technology crowd. When I prepared to leave New York in 2023 I secured a remote job in so-called “education technology.” I wasn’t supposed to be writing code but I ended up doing it anyway and moreover, my overlords paid for the then-nascent GitHub Copilot, which, God, seems so quaint now, but indeed saved me hours of on-the-clock time by autocompleting my JavaScript, a language I did not then know well, to match existing patterns in the existing codebase. I quit that job after just a couple months (obviating its purpose as a band-aid over a cross-country move) because it surrounded me with the worst people I had ever met in my life. Note this. A person who quit with me, significant because he was not one of the worst people I had ever met in my life, got me a second job in so-called “edTech,” yet again in a situation where AI-generated code made possible something I otherwise never would have done. As it turns out, that led me down a terrible and regrettable path once again crossing through the wakes of some of the worst people I have ever met. I managed to divorce myself from that second set of deplorables and I landed myself in a simpler situation, a partnership, or so I thought. That so-called “partner,” an AI poweruser, misled me, took advantage of me, expropriated money I earned, and insulted my wife. Meanwhile I undertook and underwent a number of cordial calls with emptyheaded people interested in “leveraging AI to make real change” or whatever the heck these ghouls say. You see where I am going with this — ah, so that early disgust that afflicted my beautifully sensitive friends: they were picking up the scent of the people aligned with the technology, not the technology itself, per se.

It got me thinking and observing. Let me take a page from my musical buddies; let me pick up a vibe, a whiff of spirit. I thought back to the Princeton of 2012–2016. The young smartypantses destined for the big bucks fell into two categories: investment bankers and software geeks. Polo shirts versus hoodies even then differentiated an older flavor of money from a newer. The software geeks were less predictable, most often not eschewing the old-money safety net and thus flexing their brain cells more frantically, with less decorum. But also they were the kids praised their whole lives for being smart, finally at the Mt. Olympus of smartness, their inferior bullies finally culled — the elite undergraduate experience is the best, most actualized time in your life if you are one of these geeks. So why should it ever end? Go get a job at Google; they literally call their office a “campus” and indeed you can go to the dining hall, get your laundry done, wear a backpack, and swipe into buildings with a little plastic card. You can violate your back on a beanbag chair in a brightly-painted room with whiteboards and snacks; you can let your autism show; you can use the prefix “eigen-” with aplomb.

Then ChatGPT hits the scene. If you’re one of these creatures, frozen in fat-salaried high-IQ undergraduate adolescence, you already speak fluent Reddit. You can read code even if you can’t write it, but you can probably write it. You descend on the early flashes of the AI promise like a ravenous hyena. Maybe you hunger for craven prestige and you go work at OpenAI; or you join a Effective Altruism sex cult in The Bay and you end up at Anthropic; or you embark on your journey as a “founder,” daring to describe yourself in the language reserved for George Washington and Steve Jobs but having no higher goal than to land in an incubator, just like a eggfresh chicken. Money begins to flow. A new aristocracy is born, not of D.C. or New York but of San Francisco. This new class ain’t like the old aristocrats; they haven’t been bred or trained; they don’t have that eye for luxury. It’s the aristocracy of DoorDash and bunk beds — the aristocracy that pines for no more than the college dorms. No commissioning of statues, no nights at the opera, none of that old money shit — instead, a fetish for all-night coding sessions and high-minded forum-tinged philosophy. “I’m building,” they say, ripped from any physical meaning, with mimetic pride. They aren’t, let me tell you, going out to jazz clubs to stimulate my friends’ economy in exchange for the experience of classy art.

But, you know what? Fair enough! The old-style aristocrats, the finance and real estate and MBA bros, were insufferable too, in a less naïve and thus less endearing way. The backpack’d Silicon Valley tech worker, the early adopter and truest believer in AI, the narrow-shouldered burrito-fed math whiz Kaggle superstar, won. They came out on top. But did they get high their own supply? You bet. Many such cases. The tech bros, a whole little society, a whole ecosystem that even includes some women, ascended to a priestly class on the back of gigajillions of matrix multiplications. It is their messaging which percolates through media now and it is their work that sustains the load-bearing falsehoods of the U.S. Economy. The Machiavellian question: are they ready for the throne? Can they wield the responsibility that comes with the power?

So far, the answer seems to be a big fat “no.” These people are dorks. The priestly class that shepherds the flame on which our whole economy appears to be bet is a bunch of dorks. Listen to them and look at their smirking faces in interviews filmed in liminal, YouTube-lit studio realms. God forbid, read their blog posts or their tweets. Now assume they are your overlords and you are their subject. Do you feel you are in good hands? I want my aristocrats to have some aura, man! I can handle a little subjugation, a little force beyond my control — you have to! — but does it really have to be under the thumb of a person who picks their music to be Pareto optimal, or publishes their math notes from college on Github, or, God forbid, listens to podcasts?

Why do I care, you ask? And, you ask, what about the WASP ghouls of the past, the Dulleses and the Bushes, or the spooky Rothschilds? No, I will not go out of my way to defend those freaks (though I cannot help but mention that le goût Rothschild does rock, actually, in its goofball way; and if you know the last name “Medici,” ask yourself why). But now that we are under the thumbs of the priests of Silicon Valley, we owe it to ourselves to ask what culture they might impose on us, intentionally or otherwise. Will the offline world be allowed to survive? Will you have to sign in to a twice-authenticated app to get orange juice from the store? Will that orange juice be replaced by a more-efficient synthetic composite with Pareto optimal macronutrients relative to price? What compulsory podcast will replace your kids’ textbooks? The marching band before the football game — will it give way to a DJ before a League of Legends match?

I know how silly this sounds and yes I am overdoing it. That is my right, to overdo it on my own blog. But I am serious when I worry that the preferences and culture of powerful people will create currents in the water in which you and I have to swim. Moreover, I am serious when I worry that the stunted, adolescent, aesthetically constipated culture of Silicon Valley is the culture casting those waves. But! The good thing about adolescence is that its purpose is to give way to adulthood. I want to look forward; I want to don a productive attitude. Can these boys become men? They better, lest we inherit the aftermath of Lord of the Flies played out on the Metaverse battlefield.

We begin now a new section wherein I yell advice at these dorks, having just insulted them, realizing that no chain of links would ever lead one of them to my blog. But, you never know… maybe one of them went to the Stanford Jazz Camp or whatever or found me because of my SuperCollider advocacy. Listen up, buddy, then go preach secondhand! (Then I promise to turn from sociological concerns back to technological ones, that is to say, I will write pithy statements about AI itself.)

Dear backpack’d false bay priest,

Believe me when I say that I understand you. As a kid I too straddled the metamorphosis from “early Internet” to what it is we have now. I like videogames better than movies and YouTube better than TV. I have drunk Soylent. I know linear algebra and I have pip installed my fair share of packages I will never use.

It is time to grow up. Money is fake. View counts and impressions are fake. Their first- and second-derivatives are even faker. But this world is real and so are some of the people in it. Love is real. Art is real. You are somewhere in the food chain of a new class of people, a new type of top-dog who is setting the cultural temperature of our age. At some point down the line, your time will pass, and what will be left behind will be your legacy. It is time already to start getting that right.

You are going to be rich, bro. You might be already. By definition you’ll have excess wealth. Let me try to convince you that the way you spend that is as important as the way you generate it; you will be remembered not just for your work but also your consumption — or, better yet than consumption, your commission, your creation, your patronage. Number one: get married, like actually married, and have kids. Pass down your backpack to the next generation and trade your hoodie for a suave blazer. No more roommates; buy a house. Then I want you to hire a thoughtful landscaper to make your frontyard beautiful. I want you to hire a jazz trio for the cocktail hour at your parties; I want you to throw those parties in the first place. If you download a PDF from Anna’s Archive and you like it, I want you to buy the book and put it on your shelf in your house. Buy a bookshelf that some guy made by hand. Play indie videogames. Play them with your kids. Try Rain World and Stephen’s Sausage Roll — and if you dig it, donate a grand or ten to the creators. Max out your Patreon; get on Bandcamp and play good music through luxurious speakers in your living room. If you get real big, which is to say real rich, take it up a notch: fund a night at the symphony; improve a playground at the park; dole out some commissions; how about a soccer league in your hometown?

I can overhear the voices in your ear, objecting on various grounds. Some of those objections are valid! It would indeed be more rational to skimp on the decadent expenditures and donate that money to fight malaria, or to invest it into a fund dedicated to solving the Trolley Problem once in for all. Lab-grown meat, peptides for baldness, AI Polymarket arbitrage — whatever. I hear you. What I am suggesting is not optimal and therefore not rational. I am trying to give you permission to be irrational. And look, I hear you when you zoom out to the bigger picture; marriage may be indeed an outdated, too-religious institution; you’re not ready for kids…or maybe you adopt? save a little baby’s life? Aren’t we in a new world, on the cliff’s edge of the singularity, isn’t all this old-world old-money shit going to fade away into quaint, backwards obscurity? You are the new man of a new age, commander of a platoon of AI agents, unshackled from the sclerotic and bloated culture of the past. I will not grimace even if you invoke Nietzsche! And, how about this — I’ll grant you all that even if I privately disagree, for it still renews my call to grow up. Growing up means crystallizing an opinion. Get to a place where no man stands above you; set your own agenda. In particular, don’t let rationalist forum-goblins shame you away from balling out on something cool! You will be proven right when the time comes — as long as it’s actually cool. This is the key. Now, take it from me, an outsider with sympathy, who didn’t get rich, who missed his chance to put that linear algebra to work at a DoorDash-fueled startup, but who got cool, like, halfdrunk women begging for my number after my saxophone playing alighted on the whiskey-vapors of a Manhattan bar-cool, efficiency is not cool. Quantification, benchmarking, optimization — not cool! A little excess and a little sprezzatura go a long way. And I know you can afford it!

I believe in you. You can have so much aura. Between you and me, your competition ain’t so fierce. Unshackle yourself from the past if you must, but at least dip into that past before you do. Seriously, have kids. Get some skin in the game. Children will dissolve your misplaced faith in rationality. You have all you need to be so cool. I am here if you need some guidance on how to allocate your capital towards being cool. Either way, allocate it. Startups die, technology gets replaced, trends come and go; but good memories, transmitted through the medium of aesthetic culture, will last. Big bets on the future of X, Y, Z have no guarantee to hold up, but a cool, beloved person will remain just that: cool and beloved. It’ll be worth every penny.

Your bro,

Jacob

So concludes my sociological yelling. I do think that the root of all the AI weirdness — the hype, the disgust, the confusion — is indeed sociology, the culture of Silicon Valley, of which I am not fond but which I do not view as beyond redemption. Will I get what I want — will the nerds jettison the backpacks, cut out the messianic crap, take a broader view of history and its inertia, and most of all get cool and stir the foaming economic pot of art? Sigh — no. This whole ordeal is going to play out without me steering the ship. That’s alright. What that does mean is that it’s high time to make some predictions and pull from the oven some takes on the technology itself and the economics surrounding it. If something big is happening, I want musicians to benefit and on top of that, I want us all to keep our consciences clear, so as to minimize the distractions from making beautiful works of art that will stand the test of time. Is that so much to ask?

So let us survey the territory. You can measure all people in this game by their proximity to three classes:

  • the backpack’d Silicon Valley priests, the objects of my pleading so far, in the dynamic power position, most knowledgeable about AI, most capable to use the tools, standing to gain the most from AI’s rise;
  • the old power, the holders of old capital, boomers with houses and businesses and stock portfolios and seats in Congress, not hip to the digital forefront but important because of how much weight they can throw around;
  • the underclass, the young and otherwise disempowered people outside of the tech-bro circle, the largest class by human volume but weakest in terms of actual power yet still important because they will comprise the main engine of consumption of whatever the future holds.

The first class, the priests, are competing internally to signal to the second class, the power-boomers, that the AI products they are developing are the real deal, that is to say, worth investing in. The priests are smart so they fire in parallel two guns of persuasion: fear and FOMO. “This computer program could end the world,” but also, “this computer program doesn’t need healthcare or PTO and can do the job of that 22-year old you would have otherwise been forced to hire.” The desired result is a two-stage plan: first, an injection of capital to fuel the intra-AI battles; second, the creation of a dependence on the AI tools, pulling leverage into the priests’ camp. With that leverage, they can turn a profit and lessen the dependence on up-front investment capital. The underclass hangs in the balance.

Imagine a concrete scenario: the head honcho of a law firm, say 62 years old, eyeing his own retirement and the torch-passing it will require, hears all this buzz about AI. He is in the capital-flush position to make a bet. Say he hears an interview on CNN or asks his kid about it. Surely his boomer brain explodes when ChatGPT gurgitates something resembling an amicus brief in like four seconds before his eyes. So he places the bet: now everybody at the firm has a mandate to use AI. For what? For anything, for everything! Let it rip! You can summarize and review documents faster than ever. The accountant is blazing through spreadsheets like never before! The staff software developer rebuilds a big website in a day and a half. The cost to the whole organization isn’t negligible but it’s a heck of a lot less than the salaries of three extra paralegals or another programmer.

They got him.

How can you walk back from that? Say a lazy lawyer fires off some GPTed messages without double-checking them like he promised he always would and a client gets pissed off at an inaccuracy. If the head honcho suggests going back to the pre-AI status quo, he will be met with groaning and gnashing from everyone he employs. It blows to go through spreadsheets by hand after you’ve gotten a taste; writing your own code in 2026 is like growing your own potatoes when you live next to a Ralph’s. So, the AI stays, and if anything, there might be an incentive to bump up to a more premium tier, which promises more reliability — agent review, more tokens to catch more issues before they arise. The law firm has taken a one-way path to dependence on AI. Now OpenAI can crank their prices — what are you going to do, stop using us? — and meanwhile they are training on all the data passing through their systems. The law firm has ceded its leverage; AI is an unavoidable part of the budget for now. Those with existing jobs are fine and they might even “boost productivity” (a euphemism for “work fewer hours”), but all businesses are organisms can die without fresh energy, the energy of new workers. The law school grad eyeing a job at this firm gets burned; why hire her? Why train her, why extend her patience as she builds her skills, why pay for her health insurance? Same goes for the admittedly dumb back-office jobs: the payroll person, the office manager — yes, those jobs were ultimately bullshit and ultimately easy to automate away, but they were jobs real people had, and now those real people are forced into even more competition with each other for a shrinking pie of jobs that require human beings.

I am all for automating away meaningless or undignified jobs. Nobody should have to pick cotton or schedule appointments for a dentist. But the people who would have done those things don’t just disappear! You have three options, powerf-boomers! 1) Pay people to do warmer, more human jobs — 10x the number of soccer coaches and gardeners and bakers and, yes, musicians; 2) Rev up the Universal-Basic-Income-machine and placate us with a moist welfare state; 3) Spit on the underclass until they can’t take it no more and they start killing people like you! You can guess how I rank the options.

Because of this race for capital, it doesn’t matter whether or not the AI is the real deal. The dependency will be formed and costs extracted. If there is a huge boost in so-called productivity, fine, great, but the situation of us underclasspersons will not be any different. Your job might require some performative AI use; mine might actually be made easier with the automation. Everything in between will come to pass and the situation will evolve. But the reason I wish most of all for a cooler, more enlightened new aristocracy is precisely that the coolness of the aristocracy is directly proportional to the likelihood of scenario 1 above, something like Marx’s imagined utopia rather than the grimmer reality of Marxism in practice.

OK, but is it the real deal? Is there a way to get ahead the curve on the AI thing as a principled creative person, or is the moral position total abstinence, even phyrrically so?

Enough futzing — it is time to predict.

I reiterate my earlier point: the big change is in the computer’s interface, not its output. I claim that you cannot do anything new on the computer, nothing you couldn’t in principle do before. However, what you, you personally, as a non- or semi-expert, can pull out of your computer has been vastly expanded and sped up. Final score — no revolution. I predict the fabric of society will not rip in half and we will not transition to two-day workweeks within the next decade or two. The AI tools won’t go away but also fewer jobs than you might expect will get deleted. Many many jobs are little more than babysitting anyway and the “work” done at those jobs just does not matter. I predict those will mostly stick around. But the kids, the would-be junior-level employees, are going to suffer trying to land that first job, and for no good reason. Well, the boomers are officially aging out of the workforce, and there’s a lot of 'em — will Gen Z prove to be a “nurse generation,” the caretakers of the gerontocrats in their final chapter? Charge a fair price if so… and hey, maybe Claude can run the paperwork for you.

Anyway—

Where is the AI strong? — it’s strong where you expect it to be: writing code and repeating mundane computer tasks. If you write code, and especially if you wrote code before the AI boom, you already know the deal: yes, the AI can accomplish a million times more than you can in a fraction of the time. But its code has “that smell:” nauseous comments, too much uncalled-for cleverness, a jillion files you didn’t ask for (it’s almost like the token machine has an incentive to produce more tokens!). This is fine in 99% of cases because in 99% of cases the stakes are low — the code doesn’t matter and the project itself probably doesn’t matter either — or the dependence-loop has been achieved and the AI will monitor and maintain that code in perpetuity. None of this bothers me because I feel that software people were way overpaid and way over-respected by default; I welcome them being kicked down a peg. All that worries me is the thought that we are in a late-Roman-Empire situation, on the cusp of a Dark Age. When the foundational geniuses who cast the magic spells of the first computers and the early Internet all die, how much knowledge dies with them — how vulnerable are we to a total infrastructure collapse of the Web? If you have done any Web development at all you know the whole thing is a house of cards. Ugh! Pit in my stomach! But, also, uh, not my problem.

Then there’s the even dumber stuff: the spreadsheets, the filling out of forms, the generation and delivery of reports. Yes, it will all be AI soon enough, but I expect human beings to remain at the terminal ends of those chains. An employee will, with a smile, push the button that gets the report to the employer, who will receive, if not actually read, it and acknowledge it as ‘received.’ Be chill and friendly in your office or even just on Zoom and you won’t get fired.

Some other white-collar work will undergo changes too. Research mathematics is changing (but is the way it is taught, which has far more impact, changing the same way, if at all?). Lawyers might slash their prices, but some might raise theirs if they specializing in un-fucking fucked up AI-generated legal proceedings. What I want to point out is that my list of where the big changes will happen is rather small. The more interesting question is—

Where is the AI weak? — first and foremost, creative work. Despite every attempt by the AI labs and every incentive (think of the prestige!), the art still freaking sucks. Like, it really sucks. The absolute zero that is the “good AI art” category is the dog not barking, loud as can be. The visual art you can smell from a mile away; the writing has been permanently poisoned by Reddit; and don’t even get me started on the music. If you believe that art has objective quality or value, then you can only conclude the AIs don’t have the capacity to produce it. If, on the other hand, you democratize taste and let the market decide for you what is good or not, then you likewise must conclude AI art sucks because nobody is paying for it, nobody is praising it, and nobody is developing a practice or a culture around it. A zealot will retort, “it’s just a matter of time.” Are you sure? We saw exponential explosions of progress in the math and code domains. The incentive is there for creative work, so where is the explosion? Don’t tell me it’s for lack of trying! So, to re-link-up with my friends — you are correct in your distaste for AI as it relates to creativity, and if that distaste leaks into more domains, fine. However don’t be too proud of not using AI! You don’t need to broadcast it. You get no bonus points for not making bad art, which is all you would have been able to do if you went the AI-route.

Moreover, when it comes to music and the arts, it is only a tiny fraction of any audience that has any sort of learnéd, objective taste. Most music listeners just think the musicians they listen to are cool and they subconsciously expect that coolness to rub off on them when they listen. If that sounds bleak, let me remark that they aren’t entirely wrong! Listening to Taylor Swift all day leaves an imprint on your soul, in the shape of the image Taylor Swift has created of herself. In exchange for your ear-minutes you get the glow of a safely feministic, palatable aesthetic that shapes your consumerism and your smalltalk with other Swifties. Great! Fair trade! But that means that until the AI labs go all the way to personify their computer programs as actual beings, AI-generated art and music will be at a big mimetic disadvantage. Nobody thinks ChatGPT itself is a cool person; nobody wants to have sex with Claude.

We’re on the topic of art and creativity. Is all that goes into art creative? You know the answer: of course not! This is where things get more subtle and more interesting. Consider the following situations:

  • A person working on a webcomic goes back and forth with ChatGPT asking for recommendations of what to read for inspiration based on a list of preferences. “I liked reading The Crying of Lot 49 but also Pride and Prejudice. As a kid, I loved D.C. comics, especially from the Silver Age. Diane Zhou is the best creator in webcomics today IMO. Recommend other media I might like.”

  • A musician types, “set me up a system to automate emails inquiring about live performances. Create templates where I can fill in names and other details. Also do a search for anywhere within 30 miles of Little Rock that hosts live music.”

  • An artist sets out on a big project. “I have always wanted to make a videogame using my original characters. Help me create sprites from my drawings and then a simple RPG system in Godot where I can walk around, talk to NPCs, and trigger cutscenes. I have limited coding experience.”

Do you have a problem with any of these? Why or why not? Now consider:

  • “I have uploaded all the notes on my phone into a folder. They are from years of random observations. I have also uploaded many of my personal emails. Read PLAN.md for a plot overview of a memoir I want to write and then create a first draft using my notes and emails as a style guideline.”

  • “Make a Half-Life 2 mod that lets me walk around a office building that has been overrun by aliens. Make the guns more like Halo and use the enemy behavior of Left for Dead as inspiration.”

  • “Suggest a chord progression that fits with the lyrics in SONG.md. I want sad chords, a little jazzy but not too weird, kinda like Elliot Smith.”

These seem worse, no? I draw sort of line between the first three and the last three. What is the principle underlying that line?

(If you just feel in your heart, this is a good usage and this a bad one, by all means, trust your heart, even if that means you deem all usages bad. I am digging for a principle that can act as a voice in dialogue with my heart.)

Here is my best attempt: art’s purpose is to communicate feelings and ideas. Feelings and ideas tend to be rather formless, so artistic form can shape the delivery of the feeling/idea. When we experience an artistic creation, we hope to receive the feeling or the idea at its full power. We may not be able to articulate what that idea is; in fact, this is a better outcome, for otherwise the idea wouldn’t have needed the art in the first place! But we prefer to trust that the artist worked hard on that feeling or that idea, that they fought for it, or ground it down, or grew it from a seed; a throwaway observation of an feeling or idea would feel like a waste of art. Whatever that feeling or idea is, it is not the whole of the artwork. The medium, with its own shape, delivers it, and that is something different. That formal conduit, the packaging of the art, isn’t coming straight from the heart of the artist; it exists to serve and that service may require it to recede into the background of our perceptions. For instance, I don’t want to care about the wood of the clarinets while I hear Beethoven, nor the ink on the score, nor the weather in Vienna in May 1806 even if a balmy day gave Beethoven the energy to compose an enchanted passage. Is the line between form and content always strict and obvious? No. But you see where I am going with respect to the AI question.

When you, the artist, are shooting for an artistic outcome, a package of feelings and ideas, it is healthy for you to admit that part of the formal picture needs nothing more than its basic defaults — that is to say, it doesn’t need any choices to be made. It is fine for your narratively- and visually-interesting RPG to have boring, uninspired code under the hood. When you are booking gigs, you do not need to book them “artistically;” you just need to book them! And literature review is valid by any means, whether that be the local library, Googling, or conjuring up reading lists on ChatGPT. In all these cases, the stupid, soulless, Redditbrained consensus mechanism of AI is cordoned off from the feelings and ideas you are packaging. You are still on the hook to distill and intensify the core of the art and you are still on the hook to make important choices about format — just not every choice, not down to the molecular level.

On the other hand, the second an AI’s tendrils touch that core feeling or idea…they taint it, and you have necessarily lessened your own role in your own creation. This defeats the purpose and weakens the eventual experience of your audience. Don’t delude yourself into thinking that you have a big idea and that AI will “fill in the gaps” — if there are gaps, you must fill them in yourself or choose to leave them as gaps! Nobody wants to eat Swiss cheese filled with concrete! When AI is left to generate anything deeper than the outermost layer of artistic process, it is stealing from the artist and stealing from the audience — not stealing in the way it steals and then anonymizes data, but stealing from you a better potential experience, a more direct, intense statement of creativity and craftsmanship. That is not cool. Today’s artists are right to shun those who think they can get away with that! You’re pissing in the well if you create that way and then pretend for even a second to be in the same arena as those of us who do it the right way.

I am not particularly eager to incorporate AI into my creative practices, but if and when I ever become eager, this is the test I will run. I will aim to invoke the AI either very early or very late in the process: in the research/exploration phase, or in the final roundup, putting together the cold, boring elements that nonetheless must be there for a structural reason. I do feel this approach is superior to total abstinence; because repetetive computer tasks come up in the life of an artist, AI can be helpful. But you must be real about the danger — not the danger of an apocalypse or of your job being stolen by a robot, but the danger of your soul withering from exposure to the soulless. It is a question of diet, not just exercise.

Will my principle stand? Will it make sense three generations of AI from now when the computer programs beat us at everything (or so the keepers of the programs tell us)? We shall see. I reserve the right to revise my opinion but I will resist the urge to edit this post to make myself seem more correct down the line! I promise!

As a final course after a long post, I offer some proclamations, reminders to myself if nothing else.

  • AI is a big change in how we interface with computers but not in what comes out of them.
  • If you have a job, that job will probably remain. If your job is in fact going away, either level up to be an AI-steward of the new version of your job, or find a new job, one with a human meaning. Good luck.
  • The marketing will get even more insane than it already is. Always modulate what you hear by the incentives behind the marketing. You are also free to ignore it and instead follow your own eyes and ears. Hear the dog not barking.
  • Extend some sympathy to the Silicon Valley priests. This is their moment and it is better for everyone if we pull them towards the adult world of art and culture, away from the adolescent world of efficiency and competition. That said, don’t let them get away with being dorks forever! Teach them how to be cool — maybe even for a price.
  • Many people in software got into it for the wrong reasons. They comprise a doubly reprehensible, doubly false subclass of the priests. They are all about to get owned and rightly so.
  • If you are a creative person, you are under no new threat at all. (You are under many old threats!) Keep on keepin’ on.
    • Depending on the details of your creativity, you may stand to benefit from a light dose of AI. Don’t be afraid to try if you are curious! It could be economic stuff like tax forms but you can go further, to creativity-adjacent structural tasks like coding a tool to batch-process images according to predictable rules.
    • If, however, you think even for a second that the models can offer you anything that rises to the level of meaningful art, then God help you. Don’t do it! It will suck and you will have lost the plot!
    • And if a non-creative person thinks they can prompt their way to something better than what you make by hand, let them fail. Let them embarrass themselves, but be ready to to offer a pricey contract when they suddenly need the real thing.
  • Most of all, follow your heart. Listen carefully to your what your gut tells you — it may be whispering but it’s never silent!

That’s it at the end of the day, isn’t it? Has your heart ever been wrong? My disastrous sojourn into pro-AI space came about because I thought too hard and felt not enough. Meanwhile, my artist friends kept the AI at arm’s length, possibly for the wrong logical reasons but who cares? They ended up right, uncorrupted by the slop, able to take pride in their work, that work communicating something personal and hard-earned. That is cool!