Economist: AI-written recommendation letters hurt students in at least three ways
paulnovosad · x · 2026-10-21
Economist Paul Novosad (Dartmouth) criticized using generative AI to write student recommendation letters, laying out three harms:
- Signal: it tells readers you didn't think the advisee was worth the time to write a real letter
- Vacuous content: full of genAI boilerplate that says nothing distinctive about the student
- Trust damage: it attempts to deceive the reader, creating bad vibes and distaste for the candidate
He also pushed back on "I don't care if it's AI-written, so nobody cares": a theory-of-mind failure — the equilibrium may change someday, but your advisee needs a job in the current one.
Related event: Economist Slams AI-Written Recommendation Letters as Harmful to Students(2 posts)→
More from AGI Musings
- Agents and software companies can coexist — the bar is trust, not code quality — andreisavu · 2026-10-21
- Buying software means paying for a team that decides how it improves — can agents replace it? — andreisavu · 2026-10-21
- Personalized software hinges on agents earning product-judgment trust — andreisavu · 2026-10-21
- David Patterson: by 2030 robots will make everything open 24/7 — davidpattersonx · 2026-10-21
- Citadel founder Ken Griffin says AI is "almost certainly" a bubble, but stays bullish long term — GaryMarcus · 2026-10-21
- Pigment raises $250M Series E at $2.3B valuation with NVIDIA, tops $100M ARR — mattturck · 2026-10-21