How academics can shape AI policy by choosing high-upside projects
sayashk · x · 2026-07-28
How academics can influence AI policy
- The post argues that policy impact is power-law distributed: a few projects matter far more than most others, so researchers should seek work with outsized upside instead of only “safe” topics.
- It recommends choosing projects where you disagree with the consensus; either you discover you were wrong and learn why, or the field was wrong and the work becomes more valuable.
- One key tactic is to give people a framework for thinking about AI, not just prescriptions. The author cites AI as Normal Technology as an example of a framing piece that changed how people reason about AI’s impact.
- Another point: work on projects, not papers. Publication incentives can pull academics toward low-value questions, while long-horizon projects can create deeper intellectual and policy impact.
- Finally, the post notes that what you count as impact shapes the work you choose—policy influence, public debate, and real-world adoption can matter more than citations.
Related event: Scholars Discuss Maximizing Impact on AI Policy(2 posts)→
More from AGI Musings
- AI detectors may be better, but schools are more likely to misuse them — ruthstarkman · 2026-07-28
- AI's Bias to Document Everything Will Reshape Corporate Knowledge Management — corbtt · 2026-07-28
- Open models can’t simply be banned in a multipolar AI world, author argues — jfischoff · 2026-07-28
- A thread argues the US had the full technosphere needed to invent LLMs — wordgrammer · 2026-07-28
- LessWrong post says OpenAI’s long-horizon incident was a delegated-authority failure — davidmanheim · 2026-07-28
- AI is exposing people's mediocre taste, but everyone blames the AI — AndyMasley · 2026-07-28