The Five People You Meet in AI Hell

For the last few years, working in the industry I work in, it's been impossible to avoid the hell that is The "Artificial Intelligence" Discourse. And I, dear reader, like Dante before me, would like to describe to you some of the characters I have met in this inferno.

First, there's the AI Utopian. The Utopian believes that we're on the cusp of Artificial General Intelligence and that's great. Soon, The Utopian says, we won't have to lift a finger. AI will do all the work for us while we sip Mai Tais by the pool.

Then there's The Doomerist. The Doomerist is the Utopian's emo-goth cousin. The Utopian and The Doomerist agree that Artificial General Intelligence (aka AGI aka "the kind of AI you see in sci-fi") will be here Any Minute Now. However, for The Doomerist, AGI represents "an extinction level event". AGI means the end of all human life. Perhaps all life, period, not just on Earth, but in the whole "light cone." If the The Utopian is high on the AI supply, The Doomerist is doubly so, but in a bad-trip, paranoid meltdown way.

What unites them is how they are wrong, and how wrong they are. Let me tell you, as a person who has been hands-on with AI for almost twenty years now, we're not anywhere close to AGI. Even if we were, there's a difference between intelligence-- even super-human intelligence-- and human levels of agency. Of course, the "idea guys" miss this, but there's a big fucking gap between "thinking of a thing" and "making a thing happen."

In other words, Intelligence is Not Enough:

The actual opposite pole of the discourse spectrum are the Skeptic Twins. Skeptic One thinks AI is all smoke and mirrors and that's good. Skeptic Two thinks AI is all smoke and mirrors, and that's bad.

The problem with the Skeptics is that sometimes their head is so far in the sand, their fingers so far in their ears, their yelling "lalalalala" so loud, that they missing all of the genuinely anti-social shit being generated by AI models and AI hype alike. At least Skeptic Two is wise enough to know that AI doesn't have to be real to be harmful. For example, Cryptocurrency and NFTs were the most vaporous of vaporware, and yet they enabled massive fraud, risked the whole economy, cost many jobs, and emitted a hojillion tons of carbon dioxide.

The fifth person you meet in AI Discourse Hell is The Luddite. I can have a drink with The Luddite. We're buds. The Luddite knows AI is real, and knows AI is a threat. But not in an overblown, watched-The-Matrix-too-many-times way. The Luddites are concerned about the very real risks that people are going to lose their jobs, our environment is going to get trashed, and our society is going to get stupider.

The Luddite is more right than wrong. The current crop of generative AI models are absolutely built on massive immoral exploitation of creative work. Sure, we could build new models and new policy regimes to solve this problem, but will we? As long as the tech industry and government are as cozy as they are today, there's not much incentive. And of course, I think there are some promising pro-social AI innovations-- like AI models that recognize pre-cancerous cells in medical imagery. We could also, for once in our miserable history, actually decide to listen to The Luddites and distribute the windfall fairly. We could build a society of increased leisure. But will we? Again, I wouldn't bet on it, at least short of a massive societal rupture.

So then, why is The Luddite in hell? Well, you see, between the tech industry and the governments who have zero interest in reigning it in, The Luddite is always in hell.