Debate: how did Anthropic hire all the alignment talent yet ship models 'worse aligned' than GPT?
almmaasoglu · x · 2026-09-17
A viral jab asks how Anthropic, having hired many of the top AI alignment researchers, ended up with models "substantially worse aligned than GPT." The claim sparked debate over whether alignment talent density translates into product-level alignment quality, and reflects users' divergent everyday experiences with Claude vs GPT guardrails.
More from Fun
- Every founder writes with AI but few publish: rewriting half the draft by hand — heyshrutimishra · 2026-09-18
- Greentext roasts effective altruism: from mosquito nets to galaxies of expected value — banteg · 2026-09-18
- PV Sindhu's "heartfelt" column accused of being AI-written — danish037 · 2026-09-18
- Ranking the week's takes on METR third-party evaluation, from sneers to COI claims — dgrobinson · 2026-09-18
- Polymarket puts 57% odds on AI solving the Hodge Conjecture — Polymarket · 2026-09-18
- Searching "Jevons" but Typing "jav" in a Meeting: An AI Twitter Classic — flavioAd · 2026-09-18