OpenAI's always-on agents, 700+ math manuscripts & HackerRank's AI interviewer

The panel covers OpenAI's Dots and the wave of persistent AI agents, OpenAI's release of hundreds of AI-generated math manuscripts and what it means for human mathematicians, and HackerRank's Chakra AI interviewer and its implications for fairness and evaluation. The episode closes with an explainer on mixture of experts architectures and the difference between open-weight and open-source models.