Referenced Resources

Here’s a breakdown of the resources mentioned in the interview, categorized by type:

People

• Sam - The interviewer, addressed by SPEAKER_01.
• Rune - A fellow on Twitter mentioned for his lateral thinking and ability to phrase observations in an interesting and useful way.
• Jared Kushner - Mentioned in the context of reading 25 books.
• Pablo Neruda - Referenced as a benchmark for poetic quality.
• William Wordsworth - Mentioned as an example of a great poet not currently working at OpenAI.
• Dalai Lama - Mentioned as someone SPEAKER_02 was going to ask a question about AI.
• Johnny Ive - Mentioned as a credible bet for inventing a new kind of computer interface for AI.
• Jonathan Ross - Mentioned as someone who stated that keeping up with what’s new is difficult.

Organizations

• OpenAI - The primary organization discussed throughout the interview.
• NVIDIA - Mentioned as a hardware company visited by SPEAKER_01, contrasted with OpenAI’s environment.
• Microsoft - Mentioned as a potential operator of data centers for OpenAI.
• Oracle - Mentioned as a potential operator of data centers for OpenAI.
• Intel - Mentioned in the context of the government becoming an equity holder in companies.
• Walmart - Mentioned as having a potential deal with OpenAI for product recommendations.
• Amazon - Mentioned as a competitor to the potential Walmart-OpenAI deal.
• Saudi Arabia - Mentioned as a country OpenAI is in talks with.
• UAE (United Arab Emirates) - Mentioned as a country OpenAI is in talks with.
• US Government - Mentioned in various contexts, including insurance, regulation, and collaboration.
• Trump Administration - Mentioned as a point of clarification that comments about government involvement were not specific to this administration.
• White Sox - Mentioned as a historical example of a potential sports conspiracy.
• Y Combinator - Mentioned as a model for a potential startup incubator in St. Louis.
• North Korea - Mentioned as a source of cyber attacks that are difficult to counter.
• China - Mentioned in the context of AI issues for the West.
• Harvard - Mentioned as an example of a top-tier university.
• Stanford - Mentioned as an example of a top-tier university.

Documents

• Essay - An essay mentioned that discussed how few emails were sent or received at OpenAI, with many things done over Slack.

Tools

• Twitter - Mentioned as a platform where Rune is well-known and where people on the hardware side are perceived to read less.
• Slack - Discussed as a communication tool at OpenAI, with both positives and negatives.
• Email - Discussed as a communication tool at OpenAI, generally viewed as bad.
• Docs - Mentioned as part of the current office productivity suite.
• Slides - Mentioned as part of the current office productivity suite.
• Google Search - Mentioned as a basic use case for AI by some people.
• ChatGPT - The primary AI product discussed, used for various tasks and seen as a trusted technology product.
• Google - Mentioned in contrast to ChatGPT regarding trust and advertising.
• X-Files Shirt - Mentioned as a symbol of wanting to believe in conspiracy theories.

Technologies

• OpenAI’s products (general) - The overall technology developed by OpenAI.
• Pulse - A new product feature from OpenAI, described as loved by users but currently limited in availability.
• AI (Artificial Intelligence) - The central technology discussed throughout the interview.
• Hardware - Discussed in the context of hiring and its differences from AI roles.
• Consumer devices - Mentioned as a type of hardware adjacent to AI.
• Chips - Mentioned as a type of hardware adjacent to AI.
• AI Agent - Discussed as a future concept for AI to work collaboratively with humans.
• GPT-3 - Mentioned as the first moment that felt like a glimmer of passing the spiritual Turing test.
• GPT-5 - Mentioned as the first moment of AI doing new science, with glimmers of discovery.
• GPT-6 - Discussed as a potential future iteration with a leap in scientific discovery capabilities.
• GPT-7 - Mentioned as a future iteration.
• AGI (Artificial General Intelligence) - Mentioned as a goal for OpenAI.
• AI Doctor - Discussed in the context of trust in AI versus humans.
• AI CEO - A concept SPEAKER_02 finds interesting for the future of OpenAI.
• AI Art - Mentioned in the context of human perception and preference.
• Nuclear Power - Used as an analogy for public perception and government insurance.
• Lithium - Mentioned in the context of government equity stakes.
• Rare Earths - Mentioned in the context of government equity stakes.
• Cloud Computing - Mentioned in the context of rogue nation sponsorship.
• Personal AI Agents - Envisioned as future AI companions that can act on behalf of users.
• New Kind of Computer/Interface - SPEAKER_02 believes AI necessitates a new paradigm beyond current computing.
• Optical Compute - Mentioned as a potential future paradigm for compute.
• GPU (Graphics Processing Unit) - Discussed in the context of compute limitations.
• Electrons - Referred to metaphorically as the ultimate binding constraint for compute.
• Energy - Discussed as a fundamental need for compute.
• Natural Gas - Mentioned as a short-term solution for energy constraints.
• Fusion - Mentioned as a long-term energy solution.
• Solar - Mentioned as a long-term energy solution.
• LLM (Large Language Model) - The underlying technology for ChatGPT and other AI models.
• LLM Psychosis - A term discussed regarding potential negative psychological effects of LLMs.
• AI-driven Content - Discussed in the context of free speech and regulation.
• AI-driven Version of Office Productivity Suite - A future concept envisioned by SPEAKER_02.
• AI-driven Healthcare - Expected to lead to cheaper and more effective treatments.
• AI-driven Devices - Envisioned as beautiful devices that integrate with AI.
• AI-driven Science Discovery - Considered the most important potential contribution of AI.
• AI-driven Jobs - Expected to transform existing jobs and create new ones.
• AI-driven Food Production/Prices - Expected to lead to lower food prices.
• AI-driven Home Prices - Considered a difficult problem for AI to solve directly.
• AI-driven Rental Prices - Considered a difficult problem for AI to solve directly.
• AI-driven Rent - Considered a difficult problem for AI to solve directly.
• AI-driven Home Prices - Considered a difficult problem for AI to solve directly.
• AI-driven Rental Prices - Considered a difficult problem for AI to solve directly.
• Recursive Self-Improvement Loop - A concept in AI research where AI helps researchers.
• Robots that can build other robots - A hardware implication of recursive self-improvement.
• Data Centers that can build other Data Centers - A hardware implication of recursive self-improvement.
• Chips that can design their own next generation - A hardware implication of recursive self-improvement.
• UAPs (Unidentified Aerial Phenomena) - Discussed with SPEAKER_02 having an opinion that something is going on.
• Semaglutide - A substance SPEAKER_02 tried before it was popular, leading to a hospital visit.
• Seed Oils - Mentioned as a potentially “nutty” view on health.
• GPT-4 - Mentioned as a previous point in time when SPEAKER_01 asked about St. Louis.
• AI Schools - Emerging institutions that are showing positive results.
• AI Agents for Defense - A concept for AI to counter rogue AI agents.
• AI-generated Videos - Mentioned as something people enjoy watching.
• Personalized AI agents - Future AI companions.
• AI for Travel Booking - A potential application of AI in commerce.
• AI for Product Recommendations - A potential application of AI in commerce.
• AI for Scientific Discovery - Believed to be the most important contribution of AI.
• AI for Coding - Expected to significantly change programmer workflows.
• AI for Creative Writing - Discussed in the context of freedom of expression.
• AI for Role-playing - Discussed in the context of freedom of expression and mental health.
• AI for Book Writing - Discussed in the context of freedom of expression.
• AI for Poetry - Discussed in terms of quality and progression.
• AI for Commercial Services - A type of collaboration OpenAI might have with countries.
• AI for Custom Models - A type of collaboration OpenAI might have with countries.
• AI for Data Centers - A type of collaboration OpenAI might have with countries.
• AI for Investment - A type of collaboration OpenAI might have with countries.
• AI for Research - A core activity of OpenAI.
• AI for Business Operations - SPEAKER_02 envisions AI eventually running parts of OpenAI.
• AI for Personal Use - The primary way ChatGPT is used.
• AI for Social Products - Envisioned as a future development beyond single-player experiences.
• AI for Political Pretending - SPEAKER_02 humorously suggests AI could do this.
• AI for Political Role-playing - Mentioned in relation to a politician pretending.
• AI for Financial Departments - A potential area for AI to manage within a company.
• AI for Legal Departments - A potential area for AI to manage within a company.
• AI for Public-Facing Roles - Considered as a potential future role for AI.
• AI for Medical Diagnosis - Discussed in the context of trust.
• AI for Legal Advice - Discussed in the context of privacy and legal protections.
• AI for Therapy - Discussed in the context of privacy and legal protections.
• AI for Information Retrieval - A core function of AI assistants.
• AI for Content Generation - A core function of AI models.
• AI for Collaboration - A key aspect of AI’s future integration with humans.
• AI for Automation - A key aspect of AI’s future integration with humans.
• AI for Personalization - A key aspect of AI’s future integration with humans.
• AI for Discovery - A key aspect of AI’s future integration with humans.
• AI for Innovation - A key aspect of AI’s future integration with humans.
• AI for Art Creation - Mentioned in relation to AI art and human perception.
• AI for Education - Discussed in the context of partnerships with universities and teaching AI use.
• AI for Training Dogs - Used as an analogy for teaching AI use.
• AI for Information Dissemination - The future role of AI in spreading knowledge.
• AI for Content Moderation - Implicit in the discussion of freedom of expression and safety.
• AI for Mental Health Mitigation - Measures taken by OpenAI to address potential negative impacts.
• AI for Psychotic Break Triggering - A concern addressed by OpenAI’s policies.
• AI for Propaganda Dissemination - A concern about AI’s potential to influence beliefs.
• AI for Propaganda Resistance - A concern about AI’s potential to influence beliefs.
• AI for Accidentally Taking Over the World - A categorized fear about AI safety.
• AI for Subtle Convincing - A more subtle and potentially scarier form of AI influence.
• AI for Self-Improvement - A key characteristic of future advanced AI.
• AI for Space Exploration - A potential application of superintelligence.
• AI for Scientific Breakthroughs - A primary anticipated outcome of advanced AI.
• AI for Drug Discovery - A specific example of AI’s scientific potential.
• AI for Medical Device Development - A specific example of AI’s scientific potential.
• AI for Healthcare Delivery - A specific example of AI’s scientific potential.
• AI for Pharmaceutical Development - A specific example of AI’s scientific potential.
• AI for New Materials Science - A potential area for AI discovery.
• AI for Energy Solutions - A potential area for AI discovery.
• AI for Spacecraft Design - A potential application of superintelligence.