AI amplifies intelligence, and mediocrity too

Over the weekend, a high school senior asked me a bunch of questions about AI: how much it influences my life, whether I think it's dangerous, whether it's going to replace jobs, whether we're moving too fast, and what would happen if one day we reach superintelligence.

And while I was answering, something stuck with me: we talk endlessly about what “AI will do,” almost as if it showed up on its own.

But we built it. We trained it. And on top of that, we trained it on human knowledge, human biases, human contradictions, and a not-so-small amount of human stupidity.

We basically built a machine that learned from the internet and keeps learning. What could possibly go wrong?

(When some LLM tells me about “aura farming,” we'll know it's the beginning of the end lol)

That's why humans misusing AI scare me a lot more than an AI spontaneously deciding to turn evil.

A bad decision used to affect a hundred people. Now you can automate it and get it wrong for millions. AI can amplify intelligence, but it can also amplify mediocrity with incredible efficiency.

Something similar happens with work. I don't think it's always going to be “human vs. AI.” A lot of the time it'll be someone working with AI versus someone doing exactly what they did five years ago. I have work that used to take me three weeks and now I can get done in three hours.

Now, if we use those three hours to crank out 20 PowerPoints where we used to make four, I'm not sure we're making great use of the revolution.

Maybe that's why I'm less and less interested in learning “prompt hacks” and more interested in learning to question what AI hands me. Because we already know it can be completely wrong and sound incredibly convincing. It lies to your face without consequences, and to be fair, sometimes we choose to believe it.

And yes, the technology can get a lot more powerful, and some people along with it, for better or for worse.