Great post, but I think I disagree with much of it.

Dan Carlin produced an amazing Hardcore History about slavery in the Americas. One very surprising part of the story: slavery was dying out in the American South prior to the invention of the cotton gin. With the cotton gin, though, cotton became much less expensive. This caused the demand for cotton, and for slaves to pick the cotton, to go through the roof. Instead of wearing a single wool suit for a decade until it disintegrated off of one's back, relatively normal people were able to afford lots of cotton clothing.

It seems to me that a similar dynamic (hopefully without the slaves) could play out with AI. If you dramatically lower the cost of creative work, you'll see an explosion in the consumption of creative works.

Film, TV, and music are maybe not ideal examples of what could happen, given that there's an upper limit (24 hours / day) to people's ability to consume that kind of media. We already have a ton of production of that kind of media at a cost that's relatively low.

Consider instead areas where there is just very little consumption of creative work because it is so expensive to hire a talented person to do that work.

I keep thinking about architecture: most houses are incredibly boring. But what if you could instantly generate 100 "pretty good" designs as a starting point? You'd have a much cheaper way into an interesting design. I suspect you'd still need someone skilled in architecture to put the finishing touches on your AI-imagined dream house and turn it into something structurally buildable. The skills of an architect may somewhat move from "artist" towards "structural engineer". But the job opportunities would explode, I'd think. And the opportunities in manufacturing and homebuilding to build more creatively designed homes would (should) explode.

That's just one example...

Expand full comment
Sep 21, 2022·edited Sep 21, 2022

"As the space of mature, readily-accessible generative art models expands, the financial incentive to produce probably anything other than fine art, the value of which ceased to correlate with technical skill after the invention of the camera, will evaporate."

Isn't there something just deliciously poetic about postmodern art being en mass replaced by machine-generated pieces?

Expand full comment

The year is 2040. AI models enable everyone to create their own ideal media content. Every movie adaptation of every book you’ve ever read, every dream re-casting, every half-crazed idea you thought up when you were a child, rendered in minutes.

Everyone is Stephen Spielberg, S.S. Rajamouli, and Quentin Tarantino.

The whole culture now revolves around the last scarce entertainment commodity: someone watching your stuff and commenting on it.

Finally, I am king.

Expand full comment

I see the death of hotel conference room art shows, as well as the mass unemployment of script writers and mainstream journalists.

Well, actually humanity has little to worry about. The creative will always create, and I dare say while the mundane will be consumed by the latest Hal 9000, they'll always out-create machines.

Expand full comment

Yes so many interesting references dropped in one post. I happened to be running an eLearning / AP teacher training company when Diamond Age was out and what struck me was the similarity of the learning path of the protagonist and how new doors only opened once important concepts were understood. This exactly modeled the one-room schoolhouse / home school model where multiple ages and multiple learning styles and different learning “progressions” were present and the teacher had to be aware of that. Sal Kahn was fascinated by the hysteresis loop ( common starting point in a class but various trajectories towards mastery ) inside a classroom and sought to manage that. There are seven different learning styles and yet US ( Prussian style ) classrooms use three and creative kids or kids with music, art, poetry, dance or other skills are put on Ritalin or have to leave to find private, magnet or other places to grow their skills, typically. Potentially the best thing that can happen in the US is continued pressure on public schools that comes from healthy competition.

Which brings me to generative art.

I believe that the capacity to create art is God-given and that the love for art or the creation of it that drives the process is no different than any other skill. Some people just have that extra gear and are in a whole other league but in the purest sense, my acquisition of a piece of art comes with knowing the person that made it, or having a connection with them, and my buying it or displaying hopefully gives reinforcement to the artist.

I just left a tech startup that solely focused on NFTs and generative art was used to literally create multiple, different versions of a piece of analog art for consumers. So far so good but I noticed the artists were interested from a financial gain perspective, put loose guardrails around the digital generative rules, and generally turned control over to the machine because they might get paid. Meanwhile, in the Discord channel, the buyers wanted a personal relationship with the creator, wanted them to participate with them, to understand them and all in a non-personal, cold, texting kind of way. Which just didn’t work.

So I’m thinking the best way to experience art is in the analog, the real, and the ability to create art is a God-given gift to humans to be shared with other humans in a personal way. To the extent an artist can create digital art and create a digital bond with his or her audience then that’s ok. To the extent that we think machines can do that is misguided in my opinion.

Good post MS

Just saw this for related reading:


Expand full comment

I've been fascinated by the OpenAI Chatbot results people are posting on Twitter. Reading your post just deepened my feeling of awe, hope, and a bit of fear of the new age we are entering. I don't think most of us are ready for how quickly things can change. Buckle your seat belts kids. It's going to be a wild ride!

Expand full comment

Firstly, regarding your point about AI getting calibrated on human artists' work and whether that's a procedure all writers and artists have followed, the answer is yes and here's T. S. Eliot's brilliant essay on the subject: https://www.poetryfoundation.org/articles/69400/tradition-and-the-individual-talent

Harold Bloom warmed over Eliot's ideas and made a career out of it, here's his unjustly celebrated but once-influential take on the same subject, underlining your own intuition: https://www.amazon.com/Anxiety-Influence-Theory-Poetry/dp/0195112210

Finally, here's a book my mom of all people sent me which I've been reading, and not inclined to take seriously--until I read your post: https://www.amazon.com/Return-Gods-Jonathan-Cahn/dp/1636411428

Expand full comment

Would love to hear a tech-savvy evolutionary biologist’s take on whether generative art implies cultural stasis. A deep, nerdy dive into whether the underlying mechanics allow for the emergence of genuinely new forms, or just new combinations of old ones.

Expand full comment

Fascinating post, as always. Plus extra bonus points for the Diamond Age reference. Truly #1 best sci fi of all time.

Generative models will not replace artists any more than electric motors replaced workers. Demand for human ingenuity, human talent, and skill is a bottomless pit that can never be filled.

There will always be demand for workers because we are nowhere near done with The Work. The Work is just beginning! We don't even have space colonies yet!

Generative models will not decrease auction prices for real Monets. Afterall, Banksy fetches astronomical sums even though anyone can seemingly graffiti a street corner...

Expand full comment