

Discover more from Fragmentary
Move Smart and Make Things
AI and copyright; Tom Cruise's mother; the practicalities of the day
Slightly losing my marbles from having so much on: book, other work stuff, life, radically reshaping my diet to take into account the quantified gut biome information derived from ZOE…
It turns out the new LLMs can tell you who Tom Cruise’s mother is, but can’t tell you who Mary Lee Pfeiffer’s son might be. Remind me again how you feel the ghost in this machine? I suppose, if you’re buying, you might say that that is simply an aspect of a cognition not like our own, and that insisting on the reversibility of that relationship is human-centric. But at what point does something become so unlike what we understand as thought that it no longer qualifies? Or: do we conflate thought with aliveness? Can something “think” without being alive, or even real? Those are interesting questions (if you have the patience for any of this) but I can’t really square them with the state of the art at the moment.
It also turns out there’s a list of books appropriated by AI makers for their own purposes. Look: artists always use one another’s art. No text ever stands alone. I tend to think that in any serious creative endeavour you can do quite a lot in the way of quoting, remixing, sampling, satirising, referencing, homaging before anyone should have any serious misgivings. Milton, Virgil and TS Eliot were frantic referencers and appropriators, and we not only don’t argue with it, we teach in schools how brilliant they were. I’m always amazed (and touched) when people write to me to ask to use quotes from my work which are well within fair use. Or, honestly, sometimes when they want to go a bit beyond it. And I also tend to think that for humans and art, fair use could be extended. So, in sum… I’m pretty relaxed about that kind of thing, even if I do feel an irrational territorial twinge when I see it in action (and I do think that writing into someone else’s creative universe is a weighty thing - about which, more another time). Yes, by all means: drop me a line and ask with confidence. But just as you would recognise from that jumbled paragraph that I might be less comfortable with a proposal that you just dump the entire text of my book into your for-profit MMORPG, so with LLMs.
Incidentally, if you want to dump the entire text of one of my books into an MMORPG, that’s actually a fairly simple business proposition and we can likely work something out pretty quickly. Also: I think your MMORPG sounds wonderful but I question whether it’s going to be as MM as your investors are hoping.
But LLM training? I don’t feel the same sense of creative commonality. Although I tend to see scientific and technological work as creative as much as rational, I don’t think ChatGPT is itself an artwork. I think it’s a technology which wears a simulation of creativity, and how that simulation is achieved governs its meaning. In this case, that seems to be via a straightforward systematic appropriation; yet another instance of a big tech outfit saying that the legislative environment (that is, the world everyone else inhabits and makes a living in) doesn’t suit them and restricts their power to make money and do good by… doing what, exactly? Generating a machine that can create horrible internet content and break search for all of us? Causing a strike in Hollywood? We’re all past the point where you can wave a wand and say “we created this amazing digital thing and it’s going to make everyone’s lives SO much better, so you have to let us burn the newspaper industry to the ground.” It ain’t that simple, tech bro. And if OpenAI and the rest really believed in the accessibility they think society requires from writers and artists, they’d be, well, open with their code. Which they are not.
Tell you another thing: if I believed for one second that LLMs were drifting or evolving towards consciousness, we’d be in a different conversation. I cannot easily think of a more profound project to be part of than the inception of a non-human awareness premised even partly on my texts. (I’d also be a little alarmed: I do not write texts which should necessarily be used for that. Hilarious as it would be to have an original generation of reality-mashing weirdo superintelligences, human society and the universe are not a sandpit for me to f*ck around in. That’s why we have books. Which, come to think of it, was the burden of my song all along.)
But, hype to one side, I don’t see a basis for claiming that LLMs are doing that. Nor, indeed, do the companies developing want them to; the discovery that instead of a marketable automaton they had created a trillion-dollar life form whose rights and wellbeing were now their entire responsibility would horrify them. I’d expect an immediate attempt to offload the entire project to the public purse.
So, yeah, at this point my feeling about LLM companies vs creatives is one of aching familiarity: the tech culture’s “move fast and break things” mode is out of sync. “Move smart and make things” is a better motto now.
Move Smart and Make Things
i think the last twenty years have been wave after wave of tragedy of the commons vacuum collapse in many diverse areas of human thought and commerce. turns out that a few billion dollars of VC is enough to annex a commons and enshittify the world through rent seeking in perpetuity, with maybe the occasional accidental only connect genocide thrown in as a treat. Against that backdrop AI is just another round of crowding out humans with cheaper and crappier tools.