Ontology as a Translation Layer for AI

Takeaways

• He compares magic, alchemy and AI as ways of creating, each with stuff, shape, a maker and a point.
• An ontology is the canonical definition of what is real; a taxonomy is the names a platform or person uses for it. Cloud platforms model the same ideas differently, with no direct mapping.
• In live events, "an event" is a performance to the performer, a booking to the venue and a job to the caterer, so one tool does not fit everyone.
• A good ontology works as a universal translation layer, so you need not change your backend, and he says it cuts wasted tokens mapping between tools.
• Context changes meaning: "play" means one thing in sports and another in a music app, which is where AI gets confused.