Ian Khan

The Signal: Context Scaling in Robot Policies A recent research development, RoboTTT, introduces a method to scale visuomotor context to massive lengths—up to 8,000 timesteps—without increasing inference latency. This represents a leap beyond current stat... See more

What’s Actually Happening Integrated Quantum Technologies has filed a provisional patent for MASQ—an “autonomous AI agent governance technology.” On the surface, this looks like a response to a growing pain point: as AI agents multiply, who or what ensure... See more

Signal: DenseReward – Learning from Failure at Scale A new approach called DenseReward tackles a fundamental bottleneck in robotic reinforcement learning: the lack of dense, informative reward signals. Current reward models often rely on sparse binary out... See more

The Signal: GPT-Red and Self-Play for Robustness OpenAI has introduced GPT-Red, an automated red-teaming system that leverages self-play to enhance the safety, alignment, and robustness of large language models (LLMs). Unlike traditional human-led red tea... See more

Medical imaging is the foundation of modern diagnosis. But as AI reshapes what scanners can see and what clinicians can do, the gap between technological progress and real-world delivery has never been more visible. In this episode of The Futurist, host I... See more