ChatGPT: OpenAI, Sam Altman, AI, Joe Rogan, Artificial Intelligence, Practical AI Are OpenAI’s New Agents a Step Forward or a Step Too Far?
9 snips
Aug 7, 2025 New AI agents from OpenAI are causing a stir, with some hailing them as a breakthrough while others express concern over potential risks. The discussion delves into the implications of launching open-source models and their applicability, particularly with Microsoft's integration into Windows 11. Innovations in AI technology are highlighted, showcasing how these advancements could transform user experiences. Safety and performance metrics are also examined, raising questions about the balance between progress and caution in AI development.
AI Snips
Chapters
Transcript
Episode notes
OpenAI Releases Open Models
- OpenAI released new 'open' models after a five-year gap.
- Jaeden Schafer frames this as a significant shift in model availability.
Host Promises Clarification On 'Open' Label
- Jaeden Schafer says he'll explain the difference between 'open' and 'open source'.
- He signals skepticism about calling a model fully open without shared training data.
120B Model Nearly Matches Closed Models
- The 120B model scored roughly 2600 ELO on Codeforces, close to closed models' scores.
- Jaeden Schafer highlights this as surprisingly competitive with OpenAI's proprietary variants.