Sam Altman: OpenAI CEO on GPT-4, ChatGPT, and the Future of AI
Lex Fridman Podcast #367
Sam Altman, the CEO of OpenAI, sat down with Lex Fridman to discuss the development and future of artificial intelligence (AI). In the podcast, they talked about OpenAI’s latest project, GPT-4, and how it compares to previous versions of the language model. They also touched on ChatGPT, the importance of human feedback in AI, and the potential dangers and benefits of AGI (artificial general intelligence). Here’s a detailed summary of the conversation:
The Early Days of OpenAI
Sam Altman began the conversation by reflecting on the early days of OpenAI and how the organization was mocked for its focus on AGI. He noted that when OpenAI was founded in 2015, people thought they were “batshit insane” for working on AGI. Altman highlighted the importance of being able to think big and work towards ambitious goals, even if they seem unrealistic at first.
Altman noted that progress in AI is more of a continual exponential curve than a single moment, but he believes that chat GPT, a version of the model optimized for dialogue and human interaction, was a pivotal moment for AI. The usability of chat GPT, in particular, was a significant breakthrough that made the model much more accessible and useful for people.
