Yeah, it’s this subject again. But there’s been a turn in public sentiment recently, paired in part with a political shift in the US that I hope will continue. You all know my biases anyway.

Wish I could go to this fucking cabin in the woods I built in No Man’s Sky ages ago. I’ll never go back even in the game now, man.

The promise of generative AI seems to be largely hollow these days, with clowns like Sam Altman scrambling to try and justify all the billions of dollars being poured into LLMs. Certainly there are tools that do well for certain very specific purposes (and a few of these tools may actually put me out of a job, at least in my specific field of law) but the generalized tools treated almost as magic until recently barely add any value at all to the companies that have taken them on.

I still remember arguing with people a couple of years ago about how AI “art” would never truly take off, and look, it still hasn’t — we do see it increasingly used in both amateur and professional work, but it’s still usually entirely possible to spot in many media. Use of LLMs and comparable tools in the visual arts, music, and writing all have their tells. And when they’re spotted, many if not most criticize their use, along with those who still try to justify it. I’ll acknowledge this generation of images, sound, and text through prompts does have some good uses, but again, those cases are rare and specific. I see these tools as generating more or less extremely fancy clipart, and clipart has its place and its uses, but in most contexts its use is tacky and distracting.

To take literature, LLMs are not generating bestsellers or even middling formulaic kinds of novels, at least not with such heavy human editing that a human may as well have written the thing in the first place. ChatGPT in particular loves to embellish its text with tired garbage cliches and statements that at best add absolutely nothing to the script other than getting some eyerolls from readers who have seen it all before. I’ve noticed this on YouTube lately — just explore the true crime genre for plenty of examples, most infamously JCS. All it took was one video with both artificially generated voice and script for a large part of its audience to turn on the channel forever, with old viewers even pointing out and picking at flaws in the videos that have always been present but were previously glossed over.

This is the purest form of “content”, a term I’ve always hated when used in the context of media creation. The term always felt so empty to me — consuming context like tasteless paste out of a tube, astronaut style. Then again, that’s just what these AI-heavy videos feel like. I’ve recently been watching some caving and cave diving videos, recordings of those two activities that I’d never take part in myself, and that genre of video is replete with AI-written scripts and voiceovers. It’s easy to find those tells once you know them: extreme overuse of adjectives and adverbs and overdramatization to the point of absurdity (which you might not think possible when telling the story of a guy dying in a water-filled cave nearly a thousand feet below sea level, but amazingly ChatGPT has figured out how to pull it off.)

My response is always the same once I get a whiff of that stale AI: I quit watching the video. Because if someone can’t take the time and effort to write an actual script or at the very least to seriously edit an AI-generated one, and if they instead prompt ChatGPT or whatever other LLM of choice to crap out some words they then set to an AI generated voice and post unedited, then I won’t spend my extremely limited free time watching their content.

And of course, I’ll eternally hate ChatGPT for sullying the em dash, which if you’ve been a long-time reader you’ll know is one of my favorite pieces of punctuation. Maybe it was all those Emily Dickinson poems I read in high school about death that just rubbed off on me. But fuck Sam Altman and his minions and their piece of shit tool — I’m not going to stop using it. I had it first.

Tell me when an AI system can generate text about food and convince me that it actually really wants to eat that food and maybe I’ll start giving a shit about its authenticity

That brings me to one thing I can actually gloat about a little, though in such a thirdhand kind of way that I really have no right to gloat at all, except to say that the views expressed in one of my old posts have been vindicated in a federal appellate court. Two years ago I looked at the then-recently filed lawsuit Thaler v. Perlmutter, in which computer scientist Dr. Stephen Thaler appealed the US Copyright Office’s denial of copyright protection to a visual work his AI system had generated. The Creativity Machine as he calls it is an image generator in the same family with Midjourney.

I’m very late in my latest look at the case, but it’s actually pending a writ of certiorari to the Supreme Court (fancy legal talk for they’re asking the Supreme Court to take the case) so I guess it’s still current and may even require a later update if the Court decides to take it up. This appeal stems from the DC Circuit Court of Appeals’ decision on appeal from the trial court released this March. Here’s the full text if you really want to torture yourself with reading a court decision, but the gist is that Thaler sought copyright protection for the arguable art his Creativity Machine produced after he entered prompts into it. A work needs to have an author who can actually hold that interest in copyright, however, and to this day the Copyright Office holds that author implies a human and nothing else (see the monkey selfie case that I also covered here, and where I also wrote about this guy and his lawsuit.)

Thaler’s argument isn’t that the piece in question wasn’t created by a human but rather that the human authorship requirement is outdated, a relic of the Gilded Age, which is nearly a direct quote from one of his filings. And sure, I won’t dispute that the court system still applies plenty of laws as though we were still living in the Gilded Age (especially these days, which are starting to mirror those in terms of workers’ rights, regulation, and the massive accumulation of wealth by a few.)

But the idea that only a human can hold copyright still holds, and rightly so. I argued before and still that if an animal can’t hold a copyright interest, a piece of equipment absolutely can’t, because at least the animal has independence and maybe something approaching intent. Hell, if we’re talking about certain apes, crows, and dolphins, you could maybe argue that point convincingly given the evidence that they possess self-awareness in a way the vast majority of other animals don’t.

Now for the question of whether an interest in copyright can extend to a horse girl. Special Week sure as hell appreciates food.

And that’s just the point. The court lists several reasons for and aspects of copyright law that couldn’t possibly apply to a non-living being such as lifespan, inheritance of interests, protection of property, and recognition of the intent of the author. It’s clear that the writers of the Copyright Act didn’t contemplate copyright being applied to non-human entities. And the followup argument that Thaler would properly hold the copyright interest because the Creativity Machine is his employee under the work for hire doctrine rightly fell flat on its face, the court rejecting it outright.

Finally, Thaler’s position is further eroded by the fact that the Copyright Office does grant copyright to works with human authors who use AI tools for editing or creating certain elements of said works. The issue isn’t with the use of AI in itself, then, but with the lack of connection between the user’s input and the machine’s output. I’ll continue to maintain that this lack of connection destroys intent, and without intent, you can’t call it art. As I see it, intent is what makes art, not skill or other qualities — all you need to do is be fucking creative, which some people possibly including Dr. Thaler see as embarrassing or an annoyance or something, given how much disdain they show to the efforts of working artists. And hopefully the Supreme Court also recognizes the importance of protecting the rights of actual artists against the flood of almost entirely worthless AI-generated trash.

Granted, when you have megalomaniacal freaks like Elon and Peter Thiel doing everything they can to shove AI down our throats (and also to dissolve our political rights, but that’s another conversation entirely) it may seem difficult to resist, but the court’s ruling here gives me some very small hope that at least none of the mountains of brainrot soulrot AI videos being pumped out by content mills on YouTube and TikTok will be acknowledged as copyrightable art. Not that it makes too much of a difference to most people, even the creators, who I suspect simply pump out massive volume and don’t give a damn about any of it one second after it’s published.

Anyway, these are the times we live in. I’ve been reading about ChatGPT convincing people to do insane things, and it has the charisma of a slice of plain unbuttered toast. Not exactly Her we’re talking about, otherwise I’d be talking to my own ultracomplex AI girlfriend who just happens to look and act exactly like 2B from Yoko Taro’s acclaimed sex game NieR:Automata. Fuck these AI assholes and their bullshit broken promises, but fuck them a lot more for everything they’re doing to the planet and subsequently to the human race, which after all needs the goddamn planet to operate without freezing or burning us all to death. Have fun trying to hide from that in your doomsday bunkers, you bastards.