This is a first draft. I’d like to hear where it breaks, especially which well-known AI risk you think doesn’t fit any of the three. AI safety is organized the way the people building it are organized: alignment, interpretability, evals, governance, red-teaming. That’s a supply-side taxonomy, and it makes sense...
Windows and macOS are a traditional Personal CRM, and they need hand-holding from humans. The nature of human-in-loop experience is going to dramatically change now. The context menus may look so different. Every context menu is a laundry pile of options the application developer decided on your user’s behalf, to...
There’s really no good interface for humans to perceive knowledge in multi-dimensional, non-linear relationships, on 2D screens and 1D scrolls. Today, websites allow users to randomly drop themselves on their knowledge graph and then perceive everything within a half-mile radius. But it’s still not possible to sense the global structure....