My Cousin's Kid Refuses to Use AI. He's Half Right.

Every generation of creatives says the new tool is cheating. Every generation is half right. The pattern repeats — RED cameras, Auto-Tune, CGI — and AI is the next iteration. The purists lose the technology argument and win the craft one: the tool becomes standard, taste and vision get more valuable, and a new generation walks through the door the technology held open. The working artists in the middle take a real economic hit in the process, and that pain doesn't show up in the aggregate story.

The organizations that win this transition stop paying for execution and start paying for judgment. Hire for taste, build the licensing and compensation models that let working artists participate in the new economics, and stop arguing with the tool — because the tool is going standard whether you fight it or not.

So my cousin has a son named Joshua. He's in high school, already shooting shorts on weekends, already committed to being a filmmaker. You know the kid — the one who talks about shot composition at dinner and quotes Roger Deakins at you unprompted. Good problem to have.

A few months ago his mother told me he was genuinely worked up. Not casually dismissive — philosophically opposed. Real filmmakers use real cameras, real crews, real hours in the edit bay. Anything else is a shortcut. AI makes you a fraud.

I admired it. I also disagreed with every word.

Here's the thing — I think Joshua is going to be a terrific filmmaker. I also think he's going to look back on this phase in about four years the way I look back on insisting in 1993 that real writers used pen and paper, not computers. Which is to say: the conviction was right. The target was wrong.

Because Joshua isn't actually objecting to AI. He's objecting to fakery. He's objecting to people calling themselves filmmakers who haven't done the work. Those are two different complaints, and the first one is historically illiterate — worth unpacking carefully before we go anywhere else.

Because we've seen this movie before. Several times. And every time, the purists lost the technology argument and won the craft one — meaning the tool they hated became standard, and the thing they were actually protecting (the work itself, the taste, the vision) got more valuable, not less.

In the summer of 1861, Pony Express riders galloping across the Plains kept passing the same crew over and over again — men in wagons, hauling poles and copper wire, putting up the transcontinental telegraph line. The riders waved. The crew waved back. Nobody on either side seemed to fully grasp that they were watching a job die in real time.

The Pony Express had launched in April 1860 with a marketing line worthy of a Super Bowl ad — 1,966 miles in 10 days — and a romantic operating model: 80 riders, 400 horses, relay stations every 10 miles, mochilas thrown across saddles. It is the most American story imaginable. It also lasted 18 months. On October 24, 1861, the wires were joined in Salt Lake City. Two days later, on October 26, the Pony Express was officially shut down. It had completed 308 runs, carried 34,753 letters, and lost exactly one mochila.

The riders weren't bad at their job. The job stopped existing.

And here's the part that matters for Joshua: the Pony Express became more famous after it died, not less. It got mythologized. The skill — riding 75 miles a day across hostile country, swapping horses on the move — became legend precisely because nobody had to do it for a living anymore. The craft survived as story. The economics did not.

This is half the pattern. It runs on a clock you can almost set your watch to.

In 1476, a group of Paris scribes physically attacked the printing press that German scholar Johann Heynlin had set up in their city. They smashed it. They were defending a profession — beautiful illuminated manuscripts, copied by hand over months — that was about to be eaten alive by movable type. In 1492, a German abbot named Johannes Trithemius wrote a treatise called In Praise of Scribes, arguing that monks should keep hand-copying texts because the act was spiritually superior to mechanical printing. He then had the book printed, because by 1492 nobody was reading anything important in manuscript anymore. The scribes lost. The book — the book itself, as a cultural object — won.

In 1890, the city of Lowell, Massachusetts employed a foreman, five men, and 28 boys to light its 876 gas streetlamps every evening. It was a real job, with real pride, and a real career ladder for a kid who couldn't get a factory shift. Edison's incandescent bulb did to the lamplighter what the telegraph did to the rider: erased the role, kept the function. Streets are more lit today than in 1890. There just aren't any boys with poles.

Pony Express. Scribes. Lamplighters. Carriage builders in 1905. Switchboard operators in 1965. Travel agents in 2005. Every one of them had a Joshua — somebody young and serious and convinced the new tool was a fraud and the old craft was sacred. They were partly right, in exactly the way Joshua is right. The craft was sacred. The economics around it just stopped caring.

But that's only half the story, and it's the half Joshua is anchored to. The other half — the half that actually answers his question — is what shows up after the displacement.

When photography arrived in the 1840s, portrait painters were furious. Within two generations, photography wasn't just a threat to painting — it was its own art form, with its own museums, its own canon, its own Stieglitz and Ansel Adams and Diane Arbus. Painting didn't die. A second medium was born next to it, and the work itself — the idea of the image — got bigger.

When Bob Moog shipped the first commercial synthesizer in 1964, classical musicians said it was cheating. Wendy Carlos used it to record Switched-On Bach in 1968 — Bach played on transistors — and went platinum. By the late 1970s, Kraftwerk and Giorgio Moroder were using synths to invent entirely new genres (electronic, disco, eventually techno and house) that you literally could not have made on existing instruments. The piano didn't go away. A new instrument made a new music.

The cleanest recent version is Mike Winkelmann — the artist known as Beeple. For 13 years, he’s posted a digital piece every single day on the internet, work no traditional gallery would touch because the system had no way to sell, authenticate, or care about purely digital art.

Then in March 2021, Christie's sold Everydays: The First 5000 Days — a single JPEG of all 5,000 daily pieces stitched together — as an NFT. Bidding opened at $100. It closed at $69.3 million, making Beeple the third most valuable living artist ever sold at auction. The NFT market got messy after that, but the underlying point survived: a new technology created a market for art that had no market before. Digital artists who'd been doing the work for two decades suddenly had a way to get paid like artists.

You can run the same play on YouTube (Casey Neistat, MrBeast, Marques Brownlee — auteurs of a medium that didn't exist in 2004), on game design (Journey, Disco Elysium, Inside — work that's clearly art and clearly not film), on TikTok choreography, on procedurally generated music. Every one of those is a medium that the gatekeepers of the previous medium dismissed, and a new generation walked into.

So here's the actual pattern, both halves: the new tool kills some jobs and opens some mediums. The Pony Express disappears and the telegraph operator is born. The scribe disappears and the author shows up. The portrait painter takes a hit and the photographer becomes an artist. The classical musician sneers at synths and Daft Punk happens. The traditional gallery rejects digital art and Beeple's $69 million happens.

When Joshua looks at AI and sees only the first half — the displacement, the slop, the fakery — he's correctly identifying real damage. He's also missing the second half, where the medium he hasn't named yet is being invented around him.

With that as the floor, let me steel-man what he's actually saying — because it deserves it.

Let me do his position the favor of stating it properly before I take it apart.

AI in creative work enables slop. That's not paranoia — it's observable. Social feeds are drowning in generated images. LinkedIn is a collage of bot-written thought leadership. Portfolios now include AI work that the "artist" didn't actually make in any meaningful sense. Joshua is watching a generation of kids his age claim filmmaker status because they typed a prompt into Runway and got a 10-second shot back.

He has a point about credit. If the work is entirely machine-generated and the human contribution is a sentence, calling yourself a filmmaker is a stretch. The craft tradition — the thousands of hours of lighting tests and color timing and cutting-room failures that produced a Kubrick, a Kurosawa, a Coen brothers, a Greta Gerwig — doesn't compress into a prompt. There's a real question about whether the next generation of filmmakers will have that muscle memory if the camera just does it for them.

And here's the part I take most seriously: AI is making a specific economic promise (faster, cheaper, infinite) that cuts directly against the economics that support working artists. A 2026 Rareform Audio report found 74% of artists have used AI tools professionally, 68% say their job security has declined, 61% say the value others place on their work has decreased, and 55% report lower financial compensation. Those are not imaginary numbers. They're humans with rent.

So: Joshua is right that something bad is happening. I just think he's wrong about what it is.

Every time a creative technology collapses the cost of execution, the same argument happens. The purists call it cheating. The traditionalists lose market share. And a new generation with vision but without access walks through the door the technology holds open. Here's the receipts.

In 2005, Jim Jannard — the guy who built Oakley — founded RED Digital Cinema with one goal: make 4K cinema cameras affordable. At the time, shooting in 4K was a luxury reserved for studios. A fully-loaded ARRI package ran $500,000, and wouldn't be replaced for a decade. Other 4K systems cleared $100,000 on a good day.

The RED ONE shipped in April 2007. Body-only price: $17,500.

The film establishment responded exactly the way Joshua is responding to AI. It's not real cinema. The colors are wrong. The workflow is crude. Real directors don't need this.

Four years later, in 2011, Panavision, ARRI, and Aaton announced they were ending production of analog cameras. By 2016, RED was powering over 25% of top-grossing digital films, up from 5% in 2010.

But notice what didn't happen. Roger Deakins didn't retire. Christopher Nolan didn't stop mattering. The best cinematographers in the world kept working — often with better tools. What did happen: Sean Baker shot Tangerine on an iPhone. Steven Soderbergh made Unsane on an iPhone. A generation of indie filmmakers with vision but no studio access made movies they literally could not have made before, because the gating expense — the camera — went away.

The pattern: the tool democratized. The gatekeepers yelled. The masters kept mastering. The newcomers made new work. Art got bigger, not smaller.

Dr. Andy Hildebrand released Auto-Tune through Antares Audio Technologies in 1997. Hildebrand wasn't a musician — he was a Ph.D. research engineer who'd spent his career in stochastic estimation and digital signal processing for oil exploration. The tool was literally seismic math repurposed to tune vocals. A colleague's wife joked that she'd love a device to help her sing in tune. He built it.

Cher's "Believe" in 1998 was the first prominent creative use — her producers pushed the settings to extremes, creating the warbling effect that became the Cher Effect. Cher's label tried to kill it. She refused. The track went to number one in 23 countries.

Then T-Pain's 2005 debut Rappa Ternt Sanga used Auto-Tune on nearly every song, turning the effect into an aesthetic choice rather than a correction tool. The critics called it cheating. They were right that it changed what singing meant. They were wrong that it killed music.

What it actually did: collapsed the gap between having a great song and being able to track it. An artist with taste and no perfect pitch could now finish a record. A huge amount of the best hip-hop, R&B, and pop of the last two decades — Kanye, Future, Bon Iver's 22, A Million, basically all of SoundCloud-era rap — sits on top of that technology as a deliberate creative choice.

Meanwhile, Adele still exists. Beyoncé still exists. The singers who can sing without help still get paid like singers who can sing without help. The tool didn't eliminate the craft. It expanded who got to record.

The arc from Jurassic Park in 1993 to Everything Everywhere All At Once in 2022 is the same story compressed. In 1993, putting a dinosaur on screen required Industrial Light & Magic, a crew of dozens, and the biggest special-effects budget in film history. By 2022, the Daniels made a metaphysically ambitious multiverse picture with a 5-person VFX team, a lot of After Effects, and $14 million. It won Best Picture.

The purists in 1993 said CGI would ruin cinema. They said the same thing about every subsequent generation of effects tools. The working cinematographers adapted. The practical-effects masters — people like Phil Tippett, who was the stop-motion supervisor on the original Jurassic Park — are still revered. A generation of filmmakers who would never have gotten greenlit at a studio got to make movies that would have been impossible a decade earlier.

Pull back and the pattern is consistent. Every time a creative technology collapses the cost of execution, three things happen in sequence:

One: execution stops being a business moat. The fact that you can paint photorealistic portraits, or mix a record, or color-grade a feature, or shoot 4K — stops being a business. It's still a skill, still worth having, but it's no longer a fortress. The people whose entire economic position rested on the fortress take a hit. This is real pain, and it's never distributed evenly.

Two: taste and vision get more valuable. When anyone can generate a thousand images, the question becomes which thousand, in what sequence, saying what, for whom. The editorial function compounds in value. This is why Christopher Nolan still matters even though a kid in a bedroom can render a spaceship — because Nolan knows which spaceship, in which shot, after which conversation between two characters we've spent an hour learning to care about. That's not going away. That's the thing.

Three: a new generation shows up. Kids with ideas but no access — no studio time, no agent, no connections to the Industry with a capital I. They make weird, specific, personal work the old gatekeepers wouldn't have greenlit. Some of it is trash. The best of it is the next wave. Tangerine didn't happen without the iPhone. Bon Iver's 808s didn't happen without Auto-Tune. Whatever the equivalent is for AI hasn't shown up yet, but I'd bet on it the same way I bet on open source in 2003 and everyone told me I was crazy.

Joshua, if he's reading this: your actual objection — people claiming credit for work they didn't make, a flood of low-effort slop, the devaluation of craft — is legitimate. But it's not about the tool. It's about the ethics around the tool. Photography didn't ruin painting, but the portrait painters of 1860 did take a real hit, and the industry that emerged from that shock eventually produced Impressionism because the painters had to ask what painting was for now that machines could capture likeness. AI is the machine. Film, and music, and visual art, are about to have to answer the same question.

Now — I want to be square with you about the hard part, because the narrative I've just walked you through can read as "stop worrying, everything works out." That's not what I'm saying.

Some things don't work out. Portrait painters did take a hit in the 1860s. Session musicians took a hit when drum machines got good. Letterpress operators lost their livelihoods. The aggregate story — art gets bigger — is cold comfort to the individual artist whose livelihood got repriced overnight.

The survey numbers from 2026 are stark. Among artists using AI tools, 68% feel their job security has decreased, 61% say others value their work less, and 55% report lower compensation. Only 9% of gallery professionals consider AI-generated art a legitimate medium — meaning a whole cultural institution is actively refusing to validate the new form.

The right response to that isn't to pretend the pain isn't real, and it isn't to stop the technology, because neither of those is available to us. The right response is to build the institutions — licensing frameworks, attribution standards, new union structures, compensation models that pay artists for training data, subscription models that pay artists for taste rather than execution — that let working artists participate in the new economics instead of getting flattened by them.

That work hasn't been done yet. That's where your attention should go, not to arguing with the tools.

What I actually hope Joshua does:

Keep his conviction about craft. He's right that the work matters. The hours in the edit bay, the shot composition, the color timing — those hours build the taste that will eventually separate him from every other kid with a RunwayML subscription. He should do that work. It's not wasted.

Drop the conviction about tools. His grandfather used a film camera. His mentors at film school will use digital. He will use AI. That's not a betrayal of craft — that's what craft always does. It evolves with the tools available to it. The Coen brothers used RED cameras. Soderbergh uses an iPhone. Deakins shoots digital now. None of them are less for it.

Figure out what he has that the tools don't. Taste. Stories. A point of view. An opinion about what the world is and how people inside it behave. That's the thing that will make him a filmmaker. The rest is just which buttons he presses to get the vision onto a screen.

I told him about half this at dinner. He politely disagreed. He's going to keep disagreeing for a while, because 17-year-olds with conviction don't usually fold the first time an uncle lectures them. Good. I'd be disappointed if he did.

If you've hung with me this far, let me land the plane for the actual audience of this newsletter.

If you employ creatives: The job description is changing under you. Stop paying for execution. Start paying for judgment. The designer who can hand-render a perfect layout is worth less than the designer who can tell you which of the twenty AI-generated layouts is the one that'll convert, and why. Hire for taste, point of view, and editorial instinct. Rewrite the job specs. Your team will be smaller, higher-paid, and more leveraged.

If you are a creative: The work that survives is the work with a point of view. Stop trying to out-produce the machines — you can't. Start making sharper opinions. Develop a voice that's unmistakably yours. That's the thing that can't be generated, because nobody else has it.

If you're a CEO watching your agency bill: Same advice as Wednesday's newsletter — the economics flipped, act on it. The agencies that survive will be the ones doing strategy. The ones doing execution are overpriced and on borrowed time.

If you're a parent of a kid who wants to be an artist: Get them on the tools. Don't protect them from the future because you're scared of it. The best thing you can do for them is help them build a point of view strong enough that the tools amplify it instead of drowning it.

If you're piecing together what to actually use, this week's earlier editions did the tactical work behind this argument:

The full running stack lives at theAIE.net/tools.

Amara's Law — we overestimate technology in the short run and underestimate it in the long run — is the most useful piece of advice I've got for thinking about any of this.

Short-run: people think AI is going to replace artists next Tuesday. It's not. The output is still visibly machine-shaped for a trained eye. The prompts still take taste. The editorial function still lives in humans. Don't panic.

Long-run: fifteen years from now, the way art gets made is going to look as different from today as today looks from 1995. Joshua will be 30. He'll be making work I can't currently imagine. Some of his tools won't exist yet. Some of his collaborators will be models we haven't trained yet. He'll have a point of view that came out of the exact conviction he's holding now — that craft matters, that fakery is fraud, that the work is the point. And he'll be using AI.

I'm going to buy a ticket.

I appreciate your support.