Sam Altman: OpenAI, GPT-5, Sora, Board Saga, Elon Musk, Ilya, Power & AGI

The conversation begins with Sam Altman reflecting on the chaotic board saga at OpenAI, which he describes as a 'perfect storm of weirdness.' He emphasizes the importance of having a robust governance structure and experienced board members to navigate the complexities of AI development and organizational dynamics. Altman acknowledges the board's significant power but highlights the lack of accountability to shareholders, which he views as a critical flaw. This setup frames the broader discussion on the governance of AI and the balance of power required to manage such transformative technologies effectively.

Altman then shifts to discussing the future of AI models, specifically the transition from GPT-4 to GPT-5. He anticipates that the advancements will be substantial, akin to the leap from GPT-3 to GPT-4, suggesting a rapid evolution in AI capabilities. Altman underscores the importance of the post-training step, including reinforcement learning from human feedback (RLHF), in making AI models not only powerful but also user-friendly. This segment highlights the technical and strategic considerations that drive OpenAI's development efforts.

Despite the detailed discussion, Lex did not challenge Altman's views on the nonprofit to for-profit transition strategy, which Altman heavily discourages for AI startups. This lack of pushback leaves open the question of whether there are scenarios where such a transition might be beneficial. Altman also pushes back against the notion that journalists misunderstand GPT-4's limitations, arguing that users are more sophisticated than often credited. This moment reveals a tension between public perception and the nuanced understanding of AI capabilities by those within the field.

The conversation concludes with Altman advocating for a collective governance approach to AGI, rejecting the idea of any individual having total control. He reflects on the lessons learned from OpenAI's board saga, emphasizing the need for a balance of power and robust governance structures. Altman also touches on the future of compute as a critical resource, likening its importance to that of energy. The discussion wraps up with Altman expressing optimism about humanity's collective progress and the potential for AGI to enhance scientific discovery, setting the stage for future developments in AI.

Unresolved by the end of the conversation

• Altman expressed uncertainty about the future governance structures that will effectively manage AGI, acknowledging the need for further development in this area.

What a senior practitioner would find new

• Altman suggests that AI models should allocate more compute to complex problems, indicating a shift in AI resource management strategies.
• Altman believes that a significant milestone for AGI would be a system that can significantly increase the rate of scientific discovery, setting a high bar for AGI capabilities.

Ask any question about this episode — get an answer grounded in the transcript.

Available with Premium. $9.99/month, cancel anytime.