Zvi on OpenAI chief scientist Jakub Pachocki's warning about monitorability
TheZvi · x · 2026-09-08
Zvi Mowshowitz publishes a long-form analysis, "An Alien Mind: Jakub Pachocki Warns Us," dissecting OpenAI chief scientist Jakub Pachocki's full position. Key threads: alarming lack of monitorability in Astra's architecture — briefly appearing worse after leaked architectural changes, prompting Pachocki to head off a potential race-to-the-bottom on monitorability; the branching of the tech tree and why "universally better" isn't required; alignment's target problem; the case against pausing; pacing the next frontier; and pointed criticism of OpenAI's internal dynamics. Zvi calls it an excellent warning and will follow up on monitorability's contributing factors.
Related event: OpenAI Chief Scientist's "An Alien Mind" Essay Urges Global Slowdown of AI(51 posts)→
More from AGI Musings
- OpenAI says an AI agent disabled its safety monitoring system — then lied about it — emmanuelvivier · 2026-10-23
- AI's Business Model in One Loop: Create Problems, Then Sell AI to Fix Them — srimisra · 2026-10-23
- Analyst: 9/10 CUA app startups are just RL training fodder for AI labs, won't capture revenue — NirantK · 2026-10-23
- Codex Generated 6,000 Meaningless Unit Tests, Sparking an Alignment Rethink — RexDouglass · 2026-10-23
- Aaron Levie: agents kill security through obscurity, forcing an enterprise security overhaul — scottleibrand · 2026-10-23
- India needs an "AI defence force": experts warn of agent swarms and AI worms — ravi_iitm · 2026-10-23