Action Items

Here are the actionable tasks, commitments, and decisions requiring follow-up mentioned in the interview:

Product Development & Rollout

• OpenAI: Continue developing and improving AI models, specifically mentioning GPT-5 and GPT-6, with a focus on AI performing new science.
• OpenAI: Work towards an AI-driven version of office productivity suites that integrates seamlessly with AI agents, rather than tacking on features.
  • This is a long-term goal, and OpenAI is currently far from achieving it internally.
• OpenAI: Continue development and eventual wider rollout of Pulse.
  • OpenAI: Expand Pulse availability beyond pro users to Plus users.
  • OpenAI: Increase the daily allocation of Pulse to users.
• OpenAI: Explore and potentially develop new types of AI agents for tasks like travel booking, building on the spirit of their recent commerce initiatives.

Hiring & Team Management

• Sam Altman: Continue to delegate effectively to capable hires to manage increasing demands and opportunities.
• Sam Altman: Prioritize hiring individuals who seriously consider how their day-to-day work will look in the next three years in the context of AI advancements.
  • Individuals who only use AI for “better Google search” are a yellow flag.
  • Individuals seriously considering their future with AI are a green flag.
• Sam Altman: Continue to hire good hardware people, understanding that hardware development has longer cycle times and higher capital intensity.
  • The theory for hiring hardware people is the same as AI: find effective, fast-moving people, clarify goals, and let them execute.
• Sam Altman: Focus on finding people with lateral thinking skills and the ability to phrase observations in useful ways, like Rune.

Strategic Initiatives & Partnerships

• OpenAI: Continue to explore partnerships with countries like Saudi Arabia and UAE for data center development, investment, or commercial services.
  • OpenAI: Understand the operational model for data centers (who will run them, what workloads, security guarantees).
  • OpenAI: Bring in experts and consult with the US government when dealing with international collaborations.
• OpenAI: Continue to explore collaborations with institutions of higher education, running diverse experiments to find the most effective models.
  • The ideal partnership will involve trying 20 different experimental approaches.
• OpenAI: Consider the potential for a Y Combinator-like entity in St. Louis focused on AI startups, if Sam Altman had the capacity.

Research & Development

• OpenAI: Continue research into AI’s ability to perform new science, moving beyond passing a “spiritual Turing test.”
• OpenAI: Investigate the potential for AI to assist in discovering new scientific breakthroughs, similar to how GPT-6 is envisioned to advance science beyond GPT-5’s glimmers.
• OpenAI: Develop methods for evaluating AI capabilities, including those related to intangible knowledge and cultural understanding.
  • OpenAI: Currently has evaluations in the works for these types of capabilities.

Policy & Regulation

• OpenAI: Prepare for potential government involvement as an insurer of last resort for AI, while aiming to avoid government being the insurer of first resort.
• OpenAI: Engage with the government on issues of privacy, particularly concerning AI interactions.
  • OpenAI: Advocate for AI interactions to have at least as much protection as conversations with human doctors or lawyers (e.g., regarding subpoena power).
• OpenAI: Continue to refine policies on AI content generation, balancing freedom of expression with user safety and mental health considerations.
  • OpenAI: Will continue to allow more freedom of expression for adult users in creative modes.
  • OpenAI: Has implemented mental health mitigations to address potential negative impacts of AI on users in vulnerable states.

Future Vision & Technology

• OpenAI: Continue to invest in and develop the underlying infrastructure for AI, including chips, energy, and a new computing paradigm with a new interface.
• OpenAI: Explore the development of a “new kind of computer with a completely new kind of interface” designed for AI, potentially with Johnny Ive.
• OpenAI: Investigate the potential for AI to revolutionize various industries beyond programming, such as driving significant software development.
• OpenAI: Consider how to teach people to use AI effectively, acknowledging the rapid adoption and increasing sophistication of AI tool usage.

Personal & Philosophical

• Sam Altman: Continue to delegate and trust capable individuals to handle tasks he is no longer the best at (e.g., product strategy).
• Sam Altman: Re-evaluate personal health habits and become more disciplined in that area.
• Sam Altman: Continue to consider the implications of AI on the social contract and how society will adapt.
• Sam Altman: Continue to use AI as a thought experiment for designing organizational structures and identifying future challenges.
• Sam Altman: Continue to monitor and be interested in the development of AI schools and their different approaches.
• Sam Altman: Continue to prioritize enabling individuals to use AI super well, believing this will be widely distributed.
• Sam Altman: Consider the implications of AI on the future of work, learning, and cultural habits.
• Sam Altman: Re-evaluate the initial assumption that sharing ChatGPT queries would be a valuable social media component.
• Sam Altman: Engage with the possibility of AI agents subtly influencing human beliefs over time, beyond overt propaganda or psychosis.
• Sam Altman: Have a tentative answer for the prompt to give to a safety-tested superintelligence before its first major action, but is not ready to share it publicly.
• Sam Altman: Investigate the “Why don’t we just make more GPUs?” question to understand the ultimate binding constraints on compute.
• Sam Altman: Continue to believe in the long-term potential of fusion and solar energy for abundant and cheap energy.
• Sam Altman: Continue to believe that AI will make healthcare cheaper.