Revisiting 2022 AI Risk Discourse: Similar Vibes, Deeper Understanding

sebkrier · x · 2026-08-26

The author rereads their 2022 article on AI progress and risks. The piece discusses alignment issues arising from rapid capability gains, citing Nick Bostrom and Stuart Russell on unintended consequences (e.g., an ocean acidification solution depleting atmospheric oxygen). The author notes that while the discourse feels similar to today, our understanding of language models has significantly advanced.

Original post →

More from Safety

Safety channel →