Mandy Brown, riffing on the late Joanna Russ, labels the sustained hysteria of Those Two Letters discourse as a sort of addiction.
Regular readers may recall that I’m really not keen on addiction as a diagnostic concept, and especially not in its more metaphorical deployments, such as this one; I much prefer the term habituation, and here as most other places I think it’s a far better fit. Nonetheless, Brown is right on the money when she writes:
If there is anything novel about the language of AI (and I’m not convinced there is), it’s not the mystification of capitalism or the threat of a power out of our control or the manner in which technology is used to obscure human choices and human agency. It’s that we seem to have forgotten how to talk about anything else.
(Yes, I suppose that I am also caught in that same loop of habituation—well spotted! It’s almost as if there were inescapable discursive connections to and from macroeconomic and sociotechnological trends, isn’t it? Maybe next you’d like to explain to me how there’s no ethical consumption under capitalism, or perhaps share an enlightening anecdote about a vegan of your acquaintance who also owns an iPhone?)
I spent Thursday going deaf and mad (and, I have to assume, inhaling an entire season’s worth of other people’s germs) at the Gothenburg book fair, the sort of event that sounds like the sort of place any writer would love to be, but which is in fact probably the last place in the world that anyone with so much as a scintilla of writerly sensibility would voluntarily choose to spend a day, unless there were very solid business-related reasons for doing so.
As such, I mercifully managed to miss the aftermath of the launch of Meta’s “groundbreaking” new agent, but I’ve now caught up on the second-order digests and discussions in my RSS bin, and they suggest it’s almost exactly as I would have imagined it, had I been aware that there might have been any reason to do so.
Rusty “Today in Tabs” Foster summarises the takes thusly: Muse is for losers1. Which is to say, it seems only to be capable of dealing with an extremely upper-middle-class subset of what we used to call #firstworldproblems, which it’s hard to sum up as anything more simple than “I’d rather never have to interact with any human person who earns a smaller salary than me”.
All of which links back neatly to Robin Sloan’s rather less sarky take on LLMs and agents and so forth from late last week. Sloan is one of the very few pro-AI people I can actually bear to read without feeling the urge to open my veins, and this nuanced and caveated post is a good example of why. Furthermore, however, hidden halfway through it is the obvious answer to a mystery that I hadn’t quite realised had been nagging at me for months: why can I not bring myself to care about any of this shit?
There’s a crucial limitation here, which is that your need or desire must live “inside the box”, within the realm of apps and files, pixels on screens. Most of life doesn’t happen in here. If, indeed, none of your needs or desires are app-shaped … well, that’s not unusual! I have chatted about this with several friends who pondered a while, then declared: “I don’t have any problems that software can solve.”
If that’s you, close this tab and proceed with your day.
I am shocked and delighted at how pleasurable a sentence this is to say: I just don’t have any problems that new software can solve! I invite you to try saying it yourself; perhaps you too will find it oddly liberating.
I think the last vestige of whatever personal “lifehack” solutionism I once possessed was, somewhat ironically, ablated away by becoming a user of Obsidian, and realising very quickly how easy it would become to be the sort of person who has an incredibly optimised Obsidian installation, in which they spend ten or more hours every day, to the end of working out new ways in which to optimise their Obsidian installation.
I still use Obsidian for all sort of things, as it happens! But the important evolutionary step for me was realising that I didn’t need to use it for everything, or indeed for anything. I’m almost certain that I could do even more with it, especially if I asked an LLM to help me set it up! But I just don’t know what else I’d want it to do… and as Sloan explains, if you can’t define clearly what it is that you want—if you can’t specify, to put it in design terms—then you’re just spinning your wheels and burning tokens.
I have more urgent things to do.
This meandering trail of thought eventually led me to the conclusion that “AI agents coding bespoke software for users!” is the new “mainstream adoption of Linux on the desktop!” The vast majority of people will never ask an LLM to code them a bespoke piece of software, because—to be very blunt about it—the vast majority of people can barely use the slightly-less-than-basic features of whichever office software suite their workplace provides them with, and they definitely can’t specify in the, uh, very specific sense I used above.
Admittedly, the providers of said office software have made it infuriatingly difficult to use in recent years—indeed, I find it very hard to understand why there hasn’t been some sort of global white-collar uprising prompted entirely by the widespread obligation to use Microsoft products—but that mostly just reinforces my point. The majority of folk I know with regular admin-office type jobs who talk enthusiastically about “AI” are folk who’ve found they can now just ask Excel to perform some sort of move that would otherwise have taken twenty minutes to dig out of the help files and/or a Reddit thread; the actual function is old, but being able to make it happen feels new. In these cases, “AI” is less a revolutionary technology than it is a UI band-aid draped atop the bloated trash cranked out by a seemingly unkillable software monopoly.
Those who are using LLMs and making themselves a bespoke novel-writing app, meanwhile—or bolting a natural-language interface to their personal archive of notes, or whatever else it may be—are reportedly getting great value out of these things! However, the part of me that is currently re-reading David Edgerton wonders how long it will be before maintaining those bespoke cobbled codebases becomes as burdensome as patching Windows once a week.
Maybe it won’t, though! And that’s just fine, too; good luck to them. For my part, I’m one of those people who doesn’t have a problem that “AI” can solve—and I suspect I am a very long way from being unique in that regard.
- Not the band Muse, to be clear, for whom I have a soft spot based primarily on their having been really nice people to work for when I spent a day as a KP on one of their early tours. (Their support act for that tour, the two-bit Verve copyists Starsailor, were by contrast an absolute sack of cunts.) In truth, I haven’t really kept up with their output since Black Holes & Revelations, but I kind of admire from a distance the way that Muse have become the sort of band that even the otherwise sincere, even-handed and revisionist Quietus can’t help but hate on principle. My theory: critics hate Muse because said band discovered that not only did they really enjoy doing a sort of massively histrionic progressive rock-opera thing, but also that a fiercely loyal audience really love to watch them do it, and reliably send them enough money and album sales to do it again and again and again, with bigger pyro budgets every time. God bless them. ↩︎
Leave a Reply