Our Dystopian AI Future Isn't Skynet. It's a "For You" Algorithm Stomping on a Human Face Forever

Since the debut of ChatGPT in 2022, I’ve expressed skepticism about the world-altering power of the recently-hyped large language model (LLM) “artificial intelligence” systems, your GPT 3s and your Google Geminis and your DALL-Es and similar. Some of that has appeared on this newsletter; a good overview was published by the Boston Globe last summer. It’s not hard to be more skeptical than the norm when it comes to  the consequences of these systems. In its desperate hunger for clicks, much of the media has decided to endorse a maximalist interpretation of what AI might do. I’ve cataloged them before: disrupt every industry, eliminate poverty and need, send us off of into the stars, end death. Or, alternatively, threaten everything we’ve built, provoke dystopia, exterminate humanity. Take your pick.

As I’ve said on several occasions, though the doomer and the utopian predictions point in opposite directions, they share a basic nature: they suggest that AI will sooner or later bring an end to the status quo. Things might get way better, they might get way worse, but the ordinary grinding work of life will end. This is a very attractive notion, for a lot of people, even if we get the bad version. (Maybe especially if we get the bad version.) Post-apocalyptic stories are very popular - I’m a fan myself - because they imply a sharp break with the mundane, an end to this wearying trudge through a modern world that disappoints and disappoints. Well, I’m afraid I don’t think a moderately-clever chatbot or an image generator is going to give you that future, no matter how impressive. Instead, I think all this massive investment into AI points us towards a world where what we consume is increasingly dominated by the dictates of technology and the companies that control it. That’s already happened when it comes to choosing shows to watch next on Netflix or Twitter accounts to follow. My prediction is that the very choice of what to buy will in time be automated out from under us, with our apathetic consent, and from there, who knows? Government, bureaucracy, education… Perhaps even political or military power, at which point we might finally reach that dystopia. But for sure, our children will grow into a world where AI’s nudges feel less and less like suggestions.

This future isn’t as dramatic or as sudden as Skynet dropping the bombs. But it feels much more intuitively plausible to me and thus scarier.