The Agent Reasoning Interface: o1/o3, Claude 3, ChatGPT Canvas, Tasks, and Operator — with Karina Nguyen of OpenAI | Latent Space: The AI Engineer Podcast | Podwise
This interview podcast features Karina Nguyen, a research lead at OpenAI, discussing her work on improving user interaction with AI models like ChatGPT. The conversation covers her experience at Anthropic (where she wrote the initial code for Claude.ai), her current projects at OpenAI (including ChatGPT Canvas and Tasks), and her insights into the challenges and future of AI agent development. Specific examples include the iterative development process for Canvas, focusing on user feedback and model retraining, and the importance of behavioral design in shaping AI interactions. Listeners gain insights into the practical challenges of building and deploying AI features, the evolving role of AI in various applications, and the future potential of AI agents in streamlining tasks and improving productivity.
