AI safety's foundational papers are being read by fewer people, researcher warns
JacquesThibs · x · 2026-09-18
A researcher warns that AI safety's most important older works are read by a shrinking share of practitioners, because new papers feel shinier, seem more relevant, and jobs reward staying current. He argues that a new Anthropic report shouldn't automatically outrank foundational work you've 'been meaning to read,' and says he'll help recontextualize classic papers for today's era.
More from AGI Musings
- Your unfinished ideas hit a vendor's context window first — and can be front-run — adrianscottcom · 2026-09-18
- Google Expands AI & Economy Research Program with Nobel Laureate Aghion — heiga_zen · 2026-09-18
- Gary Marcus mocks Nvidia's rise: 'the circular economy is still thriving. For now.' — GaryMarcus · 2026-09-18
- Josh Elman goes 'Jev-pilled': next wave of consumer startups built on instant AI sorting — multiply_matrix · 2026-09-18
- VraserX backs Jensen Huang: labs aren't racing to a red self-destruct button — VraserX · 2026-09-18
- Gary Marcus: We'll Get Real Answers on the AI Bubble Within 12 Months — GaryMarcus · 2026-09-18