AI
Raising the Bar
A competent essay used to be evidence of something. Now it's evidence you can prompt a chatbot. The work didn't go away. It moved.
Before AI coding assistants, a good developer wrote at about 30 words per minute, line by line, thinking about each one. That was considered excellent. Today, if my team wrote code at that pace, they wouldn’t be employed.

I keep that in mind when I read the essays warning that AI is hollowing out education. In July, Meghan O’Rourke argued in The New York Times that AI robs students of the slow struggle with language that produces discovery. David Brooks wrote that it seduces us with “excellence without effort.” A student, John Cracraft, described campuses where curiosity is evaporating. An MIT study found that students who leaned on AI showed weaker memory and brain connectivity.
They’re right about the worry and wrong about the conclusion.
Look at what the MIT study asked students to write about. Does true loyalty require unconditional support? Do works of art have the power to change lives? Those are SAT prompts. A chatbot handles them in seconds. The study proved that if you outsource an easy assignment, you remember less of it. Of course you do. A generic five-paragraph essay was never much evidence of rigor, and now it’s evidence of nothing.
My team didn’t stop working when the boilerplate got cheap. The job changed. AI writes the obvious code, and the engineers spend their day on architecture, on debugging at scale, on deciding what to build. The expectations went up. Nobody gets credit anymore for the part a machine can do.
Writing is going the same way. Ask O’Rourke to write on one of those SAT prompts and you won’t get a competent essay. You’ll get a remarkable one, and the gap between hers and the chatbot’s is where the bar now sits. The calculator made arithmetic drills worthless and left higher math alone. Google did the same to memorizing dates. Each time, the cheap part dropped out and the hard part got harder.
That’s not comforting if you teach, because it means the assignments have to change. The bar is in revision now, in the draft that shows what you cut and why. It’s in the insight a consensus machine won’t produce. It’s in your own experience, which AI can’t import. And it’s in defending your changes to someone who can tell the difference.
A confession: I’m not a natural writer. This essay took me more than ten rounds of drafting and reshaping with AI to get right. Ten rounds felt like a lot of work to me. Ten years from now it’ll probably look lazy.