Sutskever: OpenAI board doing its mission to build AGI that benefits all
OpenAI’s surprise firing of its CEO has triggered intense debate over whether the company is prioritizing its nonprofit mission of “AGI for the benefit of humanity” or succumbing to internal power struggles and poor governance. Commenters weigh theories ranging from mundane corporate politics and misalignment over commercialization strategy to speculative claims that OpenAI may be sitting on far more capable, undisclosed models. The thread also branches into a broader argument about how close current large language models are to artificial general intelligence, and whether safety-focused leadership will mean slower productization or tighter control over advanced capabilities.
Board vs. Altman: Motives and Process
- Many see the firing as driven by concerns that the CEO was manipulative, headstrong, overly focused on profit/power, and drifting from the nonprofit “benefit humanity” mission.
- Others frame those same traits as standard CEO behavior and argue the board overreacted, acting unprofessionally and damaging trust.
- The abrupt manner (surprise Meet calls, minimal notice to leadership, no clear public explanation) is widely criticized as chaotic for an ~$80B company.
- Some call it a “coup” or power grab; others rebut that boards exist precisely to fire CEOs who diverge from mission.
AGI, Secret Breakthroughs, and Safety
- A recurring theory: there may be an internal model much more capable than GPT‑4; the board and chief scientist see it as close to AGI and wanted caution, while the CEO wanted aggressive deployment/monetization.
- Supporters point to public remarks about “pushing the veil of ignorance back,” rapid upcoming capability jumps, and talk of “superintelligence” and even “discovering new physics.”
- Many commenters are deeply skeptical: they find 2023 AGI claims “tinfoil hat” territory and highlight current models’ failures on simple tasks and abstraction benchmarks.
- Others argue that such failures often reflect tokenization/architecture choices, not a lack of “understanding,” and that capabilities are improving fast.
Mission, Governance, and Power Structure
- Several note the unusual structure: a nonprofit board controlling a for‑profit entity with major external investors, a tiny board, and members with nontraditional corporate backgrounds.
- Some think this structure successfully prevented the “salesman” from capturing the research org; others fear it creates instability and invites legal and investor backlash.
- Debate over whether the board is heroically protecting humanity or dangerously out of its depth remains unresolved.
Impact on OpenAI, Ecosystem, and Users
- Many expect reputational damage, talent exodus, and loss of developer confidence, especially around the freshly launched GPT platform.
- Others argue model development is driven by research staff, not the CEO, and the long‑term mission focus may help OpenAI even if commercialization slows.
- Some hope for more open research; others expect tighter restrictions and more “doomer / pro‑censorship” governance.