Em’s supervisor could tell a ten-page engineering progress report was personally written: many other reports arriving on their desk were AI-generated. For Em, that observation exposed a contradiction in an education increasingly organized around prompting models.

  • Assignments encourage AI use so strongly that Em experiences it as an expectation.
  • The hardest class involved prompting AI for eight hours a day. Em describes the prospect of a career spent supervising it as demoralizing.
  • Students still have to present and defend their projects. Em worries about outsourcing the thinking that would let them do that.

Em asks students to keep reading, writing and asking people questions. The essay treats independent thought as something worth protecting as more intellectual work becomes easy to automate.

Caveat: this is one student’s account, not a study measuring AI’s effect on learning.

The 278-comment thread on Hacker News adds a concrete disagreement about which parts of writing are worth automating.

What the thread adds

  • jwrallie — on writing a similar report: “the hard part was not typing the words, but realizing your thought process was superficial or incomplete and needs to be refined”.
  • cadamsdotcom — a boundary for AI-assisted drafts: “Leverage tools, sure - but know the material, make your work your own, and don’t ship your first draft.”
  • FinnLobsien — a narrower case for delegation: “Once you’ve deeply thought through everything, it’s justifiable to automate writing a ton of repetitive, slightly customized grant applications.”
  • florbnit — dissent about this particular assignment: “This is a status report on a project where you have the facts available. It’s outsourcing the physically hitting keys, and copy pasting figure numbers and charts.”

Handles are pseudonymous; HN publishes no per-comment scores. The sampled order comes from HN’s ranking; this selection is a slice of the thread, not a consensus.