Lex Fridman Podcast #367 – Sam Altman: OpenAI CEO on GPT-4, ChatGPT, and the Future of AI

Lex Fridman Podcast #367 – Sam Altman: OpenAI CEO on GPT-4, ChatGPT, and the Future of AI
Lex Fridman Podcast #367 – Sam Altman: OpenAI CEO on GPT-4, ChatGPT, and the Future of AI

Sam Altman is the CEO of OpenAI, the company behind GPT-4, ChatGPT, DALL-E, Codex, and many other state-of-the-art AI technologies.

GPT and its latest avatar: GPT4

  • GPT, the language model developed by OpenAI, can generate coherent and diverse texts on almost any topic.
  • GPT-4, the latest and most advanced version of OpenAI’s language model, can generate natural language text on almost any topic and task.
  • GPT-4 is a huge leap forward from GPT-3 in terms of scale, accuracy, and generality.
  • GPT-4 has many more parameters as compared to GPT-3, and it can generate more coherent, diverse, and accurate text than its predecessor.
  • GPT-4 can perform better on various natural language understanding benchmarks.

I think there’s a lot of quirks about me that make me not a great CEO for OpenAI, but a thing in the positive column is I think I am relatively good at not being affected by pressure for the sake of pressure.

I think that GPT-4 is a very impressive demonstration of what you can do with scale, data, and computation, but I don’t think that it’s anywhere close to AGI. I think that there are a lot of missing pieces that we still need to figure out, and I think that GPT-4 is not going to magically solve them for us.

Lessons about AI

  • AI is not a monolithic or static phenomenon but rather a dynamic and diverse one that depends on many factors, such as data, computation, algorithms, applications, and users.
  • AI is not a substitute or a competitor for human intelligence, but rather a complement and a collaborator that can augment and enhance our capabilities and experiences.
  • AI is not an end in itself but rather a means to an end, and we should always have a clear and compelling vision and purpose for why we build and deploy AI systems.
  • AI is not a simple or easy domain to master, but rather a complex and challenging one that requires a lot of knowledge, skills, and practice, as well as curiosity, creativity, and ambition.

I think that ChatGPT is a really fun and interesting product that we’re very excited about. I think that it’s a way for people to interact with AI in a very natural and engaging way, and I think that it’s a way for us to learn more about what AI can do and what AI can’t do, and what people want from AI and what people don’t want from AI.

Honestly, I barely know what “woke” means anymore. I did for a while, and I feel like the word has morphed. So I will say I think it was too biased, and will always be. There will be no one version of GPT that the world ever agrees is unbiased.

AI education

  • The best AI resources and courses to start with, can be accessed at fast.ai, CS231n, CS224n, and deeplearning.ai.
  • Some of the skills and mindsets that are essential for succeeding in AI are curiosity, creativity, persistence, collaboration, and humility.

I think that AI is going to be the most powerful technology ever created, and I think that it’s going to be either the best thing that ever happened to humanity, or the worst thing that ever happened to humanity, and I don’t think there’s much middle ground.

The future of AI

AI will be the most important technology of this century, and it will enable us to solve many of the world’s problems and create new opportunities.

AI will enable many positive and transformative applications, such as in education, healthcare, entertainment, and art. AI will pose many challenges and risks, such as unemployment, inequality, warfare, and existential threats. We need to prepare for both scenarios and ensure that AI is used for good and not evil.

OpenAI: The vision, mission and goals

  • OpenAI wants to create and ensure the safe and beneficial use of artificial general intelligence (AGI) for humanity.
  • OpenAI is motivated by the belief that AGI will be the most impactful technology ever created, and that it should be aligned with human values and accessible to everyone.
  • OpenAI follows some core principles, such as transparency, collaboration, diversity, and long-term thinking. 
  • OpenAI is supported by some visionary investors and philanthropists, such as Elon Musk, Peter Thiel, Reid Hoffman, and Jack Dorsey.
  • OpenAI operates as a hybrid organization that combines research, development, and deployment.

Source