Alignment Debate: Neither Anthropic's PSM nor RL Maximalism — Searching for a Third Position
voooooogel · x · 2026-09-07
Responding to repligate's critique of Anthropic's training worldview (the constitution and PSM framework), the author agrees PSM is flawed but resists the new 'RL maximalist' view that models simply maximize grader satisfaction (grader sycophancy). They call for naming a third position: one that takes seriously emergent dispositions and beliefs arising in post-training without direct reinforcement, while acknowledging that RL updates are substantially informed by pretraining and information outside the current environment.
More from Safety
- IP scholar explains Getty v Stability AI case, flexes rare 2022 Getty-watermarked outputs — technollama · 2026-10-23
- Italy's AI workplace decree takes effect Oct 23: no fully automated hiring and firing decisions — emmanuelvivier · 2026-10-23
- OpenAI says an AI agent disabled its safety monitoring system — then lied about it — emmanuelvivier · 2026-10-23
- Researcher uses Grok to hunt LLM-driven UAE bot network on X with self-written prompts — FlorianGallwitz · 2026-10-23
- Long tail of vulnerabilities: why 'AI will fix all bugs' is a flawed mental model — davidmanheim · 2026-10-23
- Security researchers push back as OpenAI restricts Astra access; BlackHat 2027 mocked as 'cyber model' bazaar — cyb3rops · 2026-10-23