A Human Wrote This

Probably like many of you, I’m just starting to wrap my head around AI. The default setting for people in my industry is to condemn the technology as a job-killing, creativity-smothering, doomsday technology; my unions have already gone on strike to ensure protections for actors and writers. We’re told AI consumes unconscionable amounts of water and energy to generate their outputs. That AI is going to create massive unemployment in the next couple years as entry-level, white-collar jobs are gradually whittled away to nothing. “Become a plumber!” we’re told with the same hollow joy we were told to learn to code.

Coding, by the way, is one of the jobs likely to go extinct.

At the same time, we’re also told that AI is going to generate unfathomable gains in basic research, medical technology, material sciences, efficiency, and Will-Smith-eating-spaghetti video generation. AI, we’re cheerfully told at commencement addresses in which the speaker is routinely booed, will unleash a burst of prosperity for all. If it doesn’t kill us first.

Then there are those who assure us that this frenetic AI bubble will eventually burst. That there is not enough information in existence for these models to scrape in order to achieve next-level usefulness. That, eventually, what will happen is that the training models will rely more and more on their own slop, thus baking into the models all the inaccuracies, hallucinations, and flattened-thinking that limits current models. In other words, they will, eventually, eat their own tails.

I don’t know what to believe.

I’m starting to get a sense, however, of what I don’t believe. First off, any sufficiently useful (and powerful) technology will be used. Once they figured out how to shape Doritos into taco shells, for example, there was no turning back. AI has already proven itself both useful and powerful. Because of that, even if they wanted to, the idea that governments are even capable of putting this particular genie into its particular bottle now seems impossible.

In fact, it is nations themselves who are the biggest investors in AI. This is because they recognize that whoever gets to AGI (artificial general intelligence) first develops a potentially insurmountable edge over their rivals. Which is why the AI arms race has taken on such urgency. It’s also why the few established guardrails designed to prevent AI models from, say, killing humanity in order to fulfill its mission of killing cancer in humans, are so rickety. To slow down research is to, potentially, commit national seppuku.

Significant job losses in all industries seem inevitable. The general consensus among experts is that AI will eliminate 10-15% of all jobs over the next five years. Could that possibly be true?

The last time we saw even 10% unemployment was for a brief minute during Covid. We’re still feeling the economic reverberations from that. This is likely to be both deeper in terms of losses and longer-lasting, as the economy figures out how to absorb all those workers – if absorbing those workers is even possible.

I’ve experimented with AI in my writing and find it mostly useless. We’ve all become experienced at identifying that weird style of Chat GPT text which attempts to transmogrify every thought into pseudo-profundity. (“You didn’t just make dinner – you created a culinary experience.”) Where I’ve found it useful is research. The job loss stats I got for this piece are AI-generated (and human proof-checked). I’ve also used it as an editor. A typical prompt might be, “Can you look at this piece for clarity and tone?” Its responses are sometimes quite good and sometimes quite bad. Just like with a human editor!

The reason I’m comfortable using AI in this way is because I know my own writerly voice. When AI suggests anything that alters or diminishes that voice, I politely tell it to fuck off. I worry that people growing up with this technology will be less likely to develop their own styles. That pushback we’re seeing from college grads, I suspect, has something to do with this fear. Yes, we’re worried that AI will reduce our economic prospects, but more than that, we’re worried that it will reduce our humanity.

AI’s ability to scour the totality of human knowledge and regurgitate it back to us threatens to sand away the textures of our thoughts. What we prize about ourselves and each other is our individuality. I don’t think AI is going to turn us into the Borg, but I worry that relying too heavily on it will smooth some of our uniqueness away. Our brains grow when they struggle. Eliminate the struggle, atrophy the brain. My brain is already pretty small. I don’t think it can afford additional shrinkage.

In time, I suspect humans will learn to co-exist with this technology. But I don’t know what that looks like. It’s possible our entire economic model will have to change. When AI can do everything we can do better, faster, and cheaper, it’s possible “work” and “hobby” become indistinguishable. What doesn’t seem likely to me is that AI eventually burns out. I think we’re too far along for that to be possible. I think it’s here and I think it’s here to stay. I just don’t know what that means. Nobody does.