New York Cartoons

New York Cartoons

Fool Me Twice...

An update on the AI, the shitting horse, and the tech CEOs grading their own homework.

Jason Chatfield's avatar
Jason Chatfield
Sep 11, 2026
∙ Paid

Thank you to reader David Rickert for calling my attention to this guest essay in the Times on Thursday. It really gave me hope amid the doom of the past week of AI-fueled nightmares.

For anyone in class who hasn’t done the reading, it’ll be helpful to have seen this, heard about this, and quietly lost your shit over this.

Let’s begin.

10th September, 2026
New York, NY

I know, I know. In April, I spewed 2,000 words about AI regulation and promised we’d get straight back to sketchbooks and psychotic pigeons. I’m nothing if not a man of my word, in the sense that I have once again broken it.

But hear me out: Oren Cass published an essay in Thursday morning’s Times, and I read it the way I read most news of the coming apocalypse: hunched at the drafting board with my coffee going cold at the same rate as my optimism.

Oren. You know what? I’m calling him Cass from now on. Cass sounds more better-er. Anyway, Cass is the chief economist at American Compass, a conservative think tank. I am a Liberal expat cartoonist who once lost an argument with a self-checkout machine. We are not natural allies. So, when the two of us find ourselves nodding along at the same paragraph, it’s fair to assume something, somewhere, has gone properly fucking awry.

His argument (as a bit of a TL;DR) is:
Silicon Valley spent 20 years treating the public like lab rats; the public has finally noticed, and now the AI industry has turned up asking for our trust with the confidence of a man who walked into the subway car and ripped a huge fart while wearing noise-cancelling headphones.

By the way, excuse any typos or glaring errors in the following tirade… I’m writing a bunch of this on my phone in the subway. (And the smell from Airpods guy is real bad.)

So, you might have read that Americans are revolting against data centres (the buildings that, as covered in April’s rant, guzzle the electricity of 4 million homes and send them the bill) so fiercely that support has collapsed roughly thrice as much as support for nuclear power did after Three Mile Island. Yep. We are now more spooked by the sheds where the internet does its homework than we were by an actual partial nuclear meltdown.

Fewer than 1 in 4 Americans expect AI to improve their kids’ education, their jobs, the economy, their relationships or their happiness. Those are approval numbers normally reserved for jury duty and mosquitoes. When I sounded this alarm in April, I only got about 30 likes. The apocalypse, undeterred, continued apace. So it’s nice to finally have company, (even if the company is a conservative dude at a think tank.)

The President’s contribution to the debate was to post that anyone opposing data centres wants America to end up backwards and poor; the sales technique of a man in a tweed suit pounding on your screen door at midnight, yelling “Only an idiot wouldn’t want these encyclopedias!”

Cass digs up Zuckerberg’s old motto about moving fast and breaking things, then lands the cleanest line in the essay: “WE were the things being broken.”

20 years of feeds engineered like slot machines, of executives shipping apps to our kids that they wouldn’t let near their own, and now the same postcode would like us to believe the next product, the one they cheerfully describe as the most powerful in human history, is being built with our best interests at heart.

Sure, Jan.

And this time the goose really is golden. (The goose can share a paddock with the horse. The horse, long-time readers will recall, remains out, and remains shitting.)

The Prophets Would Like More Profits Pls.

It’s worth remembering what the founders themselves say. Out loud. Into microphones. On purpose.

Anthropic’s chief executive has suggested half of all entry-level white-collar jobs could vanish within 5 years. (Remember that interview?) Musk, the man who deleted my X account after I got blocked by the President, says “Eventually, no job will be needed at all!”, and vocal-fry Altman has said “Wages for many kinds of labour will head toward zero”, then, in the same breath, promises “a future where we’re all freed up to spend time with loved ones and appreciate art.”

Cool.

Appreciate whose art, exactly, Sam? You fed the artists into the machine in Chapter 1. That’s like torching the orchard and promising everyone more time with apples. Nothing you do matches what you croak. (And that’s the common thread with everything Altman has said/done since the beginning.)

The fallback plan is a universal basic income, which Cass correctly clocks as a bribe to shut up and go away: pocket money from the people who took your job, redeemable, presumably, at the company store. He has a much better idea, which I’ll get to below, because it concerns your actual retirement, and I am saving it for after the paywall like the shameless little goose I am.

Meanwhile, as the film I reviewed in April (above) pointed out, there remains more regulation on selling a sandwich than on building the machine that 5 labs cheerfully agree might end the fucking world and all human life as we know it. Since April, precisely nothing about that has changed (except the sandwich now costs $19).


“These are the brilliant researchers actively working on the technology who casually admit they do not expect their own children to make it to high school.”

Read more.


The people building these machines openly admit they don’t fully understand how they work. They tip massive amounts of data and ‘compute’ into a digital brain, and the machine starts teaching itself things it was never programmed to do. It learns to do arithmetic, it answers advanced physics questions, and it figures out research-grade chemistry all on its own… oh, then it solves Navier-Stokes before breakfast.

My stake in this is not particularly subtle. I make my living doing the two things this technology automated first and most enthusiastically: drawing pictures and writing. Last week, it was Air Mail quietly closing another cartoon market; increasingly, its clients who used to call an illustrator are now asking a robot for “something similar” at a fraction of the price.
(You think I’m catastrophising? Read KAL or Cathy’s latest)

And Cass makes the point I’ve been shouting into the bottom of my empty scotch glass for over 3 years: The labs built themselves an exquisite, endlessly refined tool for writing code, and then, he asks, where’s the plumber’s version? Something to sort the schedule, the parts list, the invoices, the bloke ringing about his boiler at 6am? Nothing.

Speaking on behalf of cartoonists, I can report that we did get purpose-built software. It just doesn’t do our invoices. (The QuickBooks plugin for Claude is rubbish.) It does our job. Badly, and for free, which in this economy is a devastating portfolio.

You’ve seen the results. They’re the Canal Street Gucci handbags of art: convincing from 10 feet, and then you lean in, and the shoddy stitching says GUCCHI, the man in the drawing has 7 fingers, and the caption reads like it’s been written in Finnish.

Alright. I need to take a breath. My station’s coming up, and everyone’s still looking at me.

Look, I take ages to write this stuff, and I hope it’s of some use to you. If it is, I’d love you to consider throwing me six bucks, or even just a dollar a week by taking out an annual subscription. I spend most of my time working on my Substack these days, so this is my main gig now. It’s not a side hustle or a hobby. Anyway. Thank you all the same for reading. I appreciate you!

This post is for paid subscribers

Already a paid subscriber? Sign in
© 2026 Jason Chatfield · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for great culture