OpenAI and ChatGPT: Lex Fridman interviews CEO Sam Altman

Click play to LISTEN to the article below

In this video, the OHT team provides commentary on an interview between Lex Fridman and Sam Altman, the founder of OpenAI. They discuss various topics related to ChatGPT and the personalization of models. They explore the idea of feedback mechanisms and the potential for users to report, upvote, or downvote responses.

They also touch on the democratization of the alignment process and the importance of keeping alignment ahead of the capabilities of models. The conversation delves into topics like human training, reinforcement learning, and the challenges of aligning models with different values across countries.

They also discuss the partnership between Open AI and Microsoft, as well as the potential business model changes and the role of safety in AI development. Throughout the conversation, they mention various features and possibilities related to the Open AI system and its applications.

Chapters

00:00 – Intro
00:22
– GPT-4
06:41
– RLHF
15:30
– AI Safety
19:05
– Alignment vs. Capability
25:14
– System Prompts
28:00 – AI Engine Prompts
31:52 – Steerability
35:04 – Business Model
38:45
– Relationship with Microsoft

Share this Post: Facebook Twitter Pinterest Google Plus StumbleUpon Reddit RSS Email

Comments are closed.