Tech Journalism

I followed an independent tech journalist for a while because I don’t trust most big news firms to report on technology (or, really, most things) without falling in lockstep with the demands of the spectacular economy—that is, to manufacture demand because someone needs to buy all this shit to keep the lights on. And I liked this particular one since frankly I trust women in the field far more than men. Vocal fry has a disarming effect on me. But over the past few months, I noticed how she became increasingly cagey towards so-called artificial intelligence. Even in a video about how LLM firms are starting to target women, because the adoption of a technology by women is a sign of its adoption by wider society (which LLM firms had thus far failed to attain), she awkwardly gestured towards the poor average consumer who just uses LLMs to search and summarize online information, and that this is a genuine use-case which LLM firms prey upon (sure, why not?).

I like to think of pop punk band Panic! At The Disco illustrating the Hegelian principle that Minerva’s owl flies at dusk, which is to say that we might not be able to fully predict how something develops over time but, in hindsight, we can reconstruct why it developed that way and not along other paths. With regards to Panic! At The Disco, obviously there was a point at which it was just Brendon Urie cosplaying as a whole band, and the music ‘they’ put out was obviously very different from back when it was a handful of horny Mormons, but they were still bangers for a while. Death of a Bachelor was great. But, there goes that owl flying out the window. Despite having a good few singles, Pray for the Wicked sucked shit for the most part—and yet, the artistic trajectory between the two albums is obvious in hindsight. Death of a Bachelor was always going to evolve into Pray for the Wicked.

It was the same with this journalist. I still followed her because I hoped the obvious signs were just of her being an awkward tech freak who browses the internet for sixteen hours a day and, if anything, needs to interact with the topic of her research. But when recently she hosted a living Clintonite fossil, everything became clear. The interview was supposed to be about the KIDS Act, a bipartisan attempt (many such cases!) to increase surveillance and decrease freedom of speech on the internet, but dovetailed into how the AI backlash is actually the same thing as wanting to age-restrict all social media. I happen to disagree with Sanders’ proposal for public co-ownership of AI because there’s no real value to be gained, but the journalist and the interviewee disagreed not only with that but with his and AOC’s proposed moratorium on datacenter construction because the whole internet needs datacenters. And also because we need to compete with China on AI (?). And also because AI is actually a democratizing force in the information war which the government shouldn’t dare touch with its greedy hands. And also because AI can diagnose medical conditions and we shouldn't let the government access that either (what about the AI firm...?). And also because fascists (which ones?) also want to stop datacenters from being built. So weird. But the trajectory was obvious with each interview touching on AI.

This was deeply disappointing and also disturbing for me. These talking points aren’t just like liberal in the American sense. They’re liberal in the classical sense. Fucking libertarian. I feel like over the past year I’ve developed a somewhat nuanced perspective on LLMs both because I am being forced to work with them and also because, following Mao, one shouldn’t have an opinion without properly investigating it. LLMs are perhaps most useful for formal analysis of datasets, to understand what we might call grammatical structures. They are also mildly useful for software development, with the caveat that they are quite bad at “computer science” proper (since they can generate, sometimes, the correct syntax to approach a problem, but are terrible at interpreting design patterns and informal—dare I say creative—aspects of the science). Perhaps those use-cases offer some value, but not enough to offset their ridiculous social costs. And yet, those costs may not be intrinsic to LLMs themselves, because China was able to develop their own LLM at a fraction of ours. This suggests that AI is not being developed with a view towards its actual usefulness, but to aggressively expand an economy which has run out of other prospects, and to colonize the general intellect which up to now has mostly resisted commodification. I’m especially suspicious of the role played here by fossil fuel firms since, at a time when clean energy is starting to become cheaper and more feasible, we boosted the shit out of energy demand (and again, unnecessarily so; see China) as if to cushion the blow.

Sorry for the rant. Just frustrated! Think critically, babies. You can't abstract a thing from the social and material infrastructure which makes it possible. Maybe in a better world, we can have Dr. ChatGPT and we can work three-hour days and we can have ice cream for dinner. I don't think anyone's a bad person who uses AI right now, either. It's not a moral thing. You just can't talk about what value AI offers when the cost is fucking over working communities and sucking the planet dry.