This morning I had a Teams meeting with a coworker. He’s an incredibly bright kid with whom I’ve worked directly for the last five years or so. He was trying to use a tool I wrote, and running into trouble with a feature I just added. Over the prior two weeks, he’d been in two meetings where I had explained this feature, and he had a sample file, written in JSON, in hand that had usage examples for the feature. When we looked at it together it was like he was having trouble seeing the example amidst the rest of the JSON content.
Never mind that I’m starting with an anecdote rather than hard data, there are six million other caveats here. Maybe he wasn’t paying attention in the meetings (though he actually asked for the first one). Maybe I am bad at explaining things (I am). There’s a host of inherent problems when you are trying to point out something systemic. Ofttimes the people who are not affected by the thing can’t see it. Always the people who are profiting from the systemic issue don’t want it seen and the evidence for it can be refuted by naming a hundred other possible causes. I’m sure all of these phenomena have been observed and named. I could go to AI to ask about them, but I don’t think it would make my writing better.
(Also, J., if you end up reading this: I still love you and I still think you’re incredibly smart. It’s only because you are such a normally sharp mind that I noticed something wasn’t right).
The other day on (I think) this platform I read a brief note from a non-fiction writer (I now can’t find it, so if this was you and you read this, please link to your post in the comments). What is happening to him is this: people are going to AI and getting them to summarize one of his books. Based on that summary, they develop some objection to one of his ideas and then go to post that objection on his Substack. Here is what he said about it: “These people think they are cyborgs, and their minds are gone.”
In the 1940s and ‘50s, cigarette companies started running ads featuring doctors; “More Doctors Smoke Camels,” and all that. It was that period in America where everybody knew that smoking was bad for human beings, but for various reasons they either didn’t want to know it or they didn’t want it known. And since hard data hadn’t yet been collected, there was a space in which you could subtly craft counter messaging. “Hey, doctors like smoking. And they know about health, right? So it can’t be that bad for you.”
The peddlers of AI came to their “More Doctors Smoke Camels” messaging almost immediately. They had to, because everybody who was working with, in, or anywhere near the education of young human beings knows that AI absolutely destroyed the ability of anyone to do that education. So there was that notion of cyborgs — that somehow getting AI to summarize books for you was going to make you smart, that AI “frees your brain to do higher order work”, or that Claude could make you a super coder1. They more or less dropped the messaging when Corey Doctorow utterly demolished the idea, coining the notion of Reverse Centaurs. Now they tout the small percentage increases in the abilities of new models while financing them as if their growth were exponential.
If anything of this messaging remains, it’s the idea that, for educated adult humans, there’s some set of ways to use AI that is not harmful to their brains. And there well may be, but the ways that it is actively harmful to the human brain are utterly overwhelming them. Moreover, I don’t think there’s anybody out there reading this who doesn’t already know it. It’s that we’re still in some liminal state where everybody knows it, but we’re acting as if it’s not true2. Instead, it’s as if in the 1940s and 50s, instead of running ads about what doctors smoked, we just went ahead and leveraged our entire economy on the future of tobacco industry profits3.
Again, I’m not saying anything new here. Everybody knows — even if they don’t know they know it — that AI is terrible for the human brain. What I’m here to say is that I can see it happening to you. Yes, you — the adult who perhaps thinks you’re getting more done because AI is summarizing your emails and meetings and essays and non-fiction books and writing your responses back for you.
I can see you having trouble engaging with the ideas I’m presenting when we’re at work, things you used to absorb with ease. I can see that you didn’t absorb anything beyond the barest outline of my last message, and that none of its actual content made its way into your mind.
I can see that you can’t read anymore. I can see that you can’t write.
I can see that you are forgetting how to actually engage with another human, face to face, in dialog. Even through the scourge of social media, you used to still be able to look up from your infinite scroll and do that for a minute or two, at least until the phone buzzed again. Your mind is deteriorating, right now in front of me, and it’s worse every day. You don’t seem to notice it happening. You are a healthy, adult human being and you are in visible, cognitive decline. And it scares the hell out of me.
