When you ask “Chad” (as we refer to Chat GPT, or any LLM) to make an image, a video, or write you a story or something, this video is a great approximation of what’s happening behind the curtain. Kutiman had LLMs figured out 16 years ago, although LLMs didn’t exist back then, and he had no inkling what was coming down the pike. Then again, it doesn’t matter why someone is right. There are no style points in right/wrong questions. Kutiman took bits and pieces of the internet, and made something transformational out of them.
There have been plenty of lawsuits filed against LLM outfits over copyright and trademark infringements. They’re all losers, at least to date. Don’t be misled by a ginormous settlement by Facebook for $1.5 billion or so, paid to a gaggle of authors. That really wasn’t a lawsuit based on LLMs scraping the internet and re-using what they found in amalgamated forms. Facebook got caught using libraries of pirated books to train their bots, so they lost the case. Legally accessible stuff has never gotten a plaintiff a copyright win against any Chad that I know of, probably won’t ever, and shouldn’t, at least in today’s legal landscape.
The Fair Use Doctrine comes into play here. Fair Use is a legal principle in U.S. copyright law. It allows people to use copyrighted material without permission in certain situations — typically for purposes like criticism, commentary, news reporting, scholarship, or research. It’s easier to skate on Fair Use if you’re futzing around with factual stuff, and/or the output is educational. Using only modest amounts of any particular pillaged work helps. If the final product isn’t likely to harm the value of the original work, you’re probably golden. But the key detail is that the output has to be transformative.
So what qualifies as transformative, you might ask? Well, I am not a law talking guy, so I’ll keep it as simple as I am. If you transmogrify the original work to say or do something new, add commentary, change its purpose, or make art out of the artless mess you found, you’ve transformed it. You’re probably in the clear, legally. Although if Disney sues you, being technically correct is no longer the best kind of correct. The kind of correct with 400 lawyers on speed dial is the best kind of correct these days. Remember, the courts can stay irrational longer than you can stay solvent. Facebook found that billion-and-a-half in their couch cushions. I doubt you can pull off that trick.
Regular humans don’t produce anything artistic or intellectual in a vacuum. People are just like LLMs. Everyone is magpie for information their whole life long. You might get your information in the library, or by watching Two Broke Girls, but whatever you know, someone else has already known it, unless you’re John Von Neumann, and he’s dead. Or maybe Tarzan. But even he picked up stuff from the apes, including fleas. Everyone spends their whole life integrating disparate, already experienced things into their daily activities. Intellectual work is no different than shoveling the driveway. You don’t invent a shovel before putting on your mukluks and heading out. And all your curses have been uttered before, believe me. By me for certain, and probably by several other people who slipped on the ice.
So now LLMs can synthesize all sorts of inputs and produce all kinds of output much faster than homo sapiens, or whatever species they hire to write the New York Times. I really can’t keep up with the pace of it. Here’s a video produced with a suite of AI tools from about five months ago:
There’s a heaping helping of uncanny valley in there, but the days of putting 13 fingers on two left hands is pretty much over. Homages, riffs, and little bits of ephemera are giving way to full-blown productions. That was inevitable. It’s a cinematic Rubicon, the same sort of paradigm shift that self-publishing books produced. I never would have been able to publish a book unless I could pull the self-publishing end-around on nineteen women named Heather at Simon & Scribner Random Penguinstein and Co., all chanting Boxwine Über Alles while typing my rejection letters. And, no, “the end-around” is not a euphemism, although it has plenty of potential.
Just a month after that last one, someone else posted a short Sci-Fi film that looks miles more detailed:
LYRA is a haunting, beautiful glimpse into a future where humanity is gone, and a lone robot walks the Earth. Prompted and produced by Top Notch Cinema, this 3-minute cinematic short was created using a suite of cutting-edge AI tools including MidJourney, SeedDance, 11Labs, Adobe Firefly, and Kling. This film explores themes of identity, memory, and what remains when everything else is gone. As AI tools evolve, we’re just beginning to see how creators will harness this power to tell deeper, more meaningful stories. This is just one example of what’s possible.
If you’re one of the nice folks hoping that any obvious synthetic qualities of AI-generated video and audio will eventually turn people off, maybe you should review the way the music industry adopted pitch-correction software to make singers out of mumblers. Pitch correction was designed to be undetectable, a Lilliputian fraud on the audience, but it took about fifteen seconds for the talent-challenged to turn it up to eleven, where the effect became obvious.
That example was from 27 years ago, by the way. Everything on pop stations has that robotic twang now. People prefer it, I guess. I think it’s like eating Hamburger Helper without the hamburger, but facts is facts.
Back to the videos. Of course 99% of AI video things are bound to be science fiction, or sparkly vampire bodice rippers. The people who produce them are the biggest consumers of the genres. They are bound to get high on their own supply. But they’re going to have a hell of a time trying to “tell deeper, more meaningful stories,” when they’ve never read one, never mind told one.
It’s a Director’s Cut generation. They will watch endless hours of “cinematic universes” filled with creatures named Glorp or Flapdiddle wandering the actual universe in search of new galaxies and a plot. I have no idea how many version of Star Wars are out there, but at this point I imagine you’re watching droids doing their laundry and defrosting their refrigerators. Hell, even the original Star Wars barely had a plot.
That what Kutiman brought to the table, and why his thang was different than most. He was the first true prompt engineer. Each of the videos in his concatenation was A Thing on its own. He gathered them up, painstakingly I’m sure, because finding particular stuff on the internet sixteen years ago was no picnic. Then he used his own artistic and comedic sense to make something more entertaining than the sum of its parts. His work was transformative. It was His Thing.
Where’s it all going? Probably nowhere very satisfying. The amalgamated output of pablum is more pablum. Good writing is not to be found very often on the internet, or even on the best seller list. You certainly won’t discover its secrets on things like the Hemingway app. And I hate to break it to you, if you need that thing, you’ll never be a good writer.
So Chad is getting really powerful. I know it reads some good writing, because its bots beat the hell out of this website, for instance. But it has no idea what to make of it, and neither do internauts using various Chads to make videos. I explained it to Chad a long time ago, but he didn’t listen. Neither did anyone else:
Granpa told me all about the genie in the lamp.
It’s the oldest story ever and came from the land of the sand and the women with only eyes. It’s in there, the genie of everything, but you have to find him and let him out. Then he’s out and you have to figure what to do with him. Granpa says he’s wonderful but as dumb as a stump, just like all of us. He can do anything but doesn’t know what to do. He needs guidin’. — A Thousand and One
Take that, Kutiman. I wrote that one 18 years ago.
One Response
I could’ve gone the rest of my days without being reminded of “Gypsies, Tramps and Thieves”. Fortunately, there are numerous effective antidotes, so I should be able to drive it from my head without resorting to inductive phrenology.
As far a I know, I haven’t listened to anything using pitch correction (I didn’t listen to Cher, just now). I’m probably wrong about that, but I don’t listen to any modern pop, so the odds are pretty good. I have listened to singers who needed it.
As for AI, it’s a tool, and can be used for good, or slop. Mostly, we’re seeing laziness win out. I’m not hopeful that will change. Fortunately, there is human-generated content a-plenty out there, sufficient for my consumptive needs for the rest of my days.