
Real-time AI is changing the kinds of products we can build. Yesterday, I was invited to an event yesterday with IEQ Capital to hear from @_bschmidtchen (CTO, Reactor) and Dashiell Flynn (CTO, Cleveland Cavaliers) speak about the future of AI. A few topics that came up were: 1. The future of media is going to adapt while you watch it, 2. digital characters respond as you interact, and 3. robots will use nearby compute that are going to be a part of telecom towers to react to their surroundings. But, those experiences depend on inference that is fast enough to keep up with the interaction. The Cav's perspective is that the opportunity is in turning player tracking data into useful decisions: helping coaches evaluate situations, exploring alternative actions, and finding small improvements that matter in competition. The most important concepts that came out of it were that: • The right response time depends on what the system needs to help someone do. • As basic AI generation gets cheaper, the products and services built around it become more important. • Increasingly realistic AI experiences need traceability and clear signals about what is synthetic. For my own work in cognition and robotics, this connects to a question I keep returning to: How do we build systems that observe a changing person or environment, combine multiple signals, and respond usefully as things unfold? That could mean a robot adapting to someone’s movement, a training environment responding to behavior, or a tool helping someone understand their performance. This is where I think the true future of AI is going: turning observations into timely, useful action, then learning from what happens next. We also both have really cool jackets. #ai #neurotech #robotics


















