OK, I’m going on record 🙂
The future of AI will eventually be on-device compute, not cloud-based server farms. Improved programming techniques and growing raw hardware power will make this possible out of necessity and economics. And Apple is best-positioned to facilitate this. Not because they are better programmers, but because they have better silicon.
And once Apple finally embraces AI, the sea-change everyone’s been talking about for 4 years will finally happen — because it’ll truly be in everyone’s pocket.
America will lose the AI race, because we’ll have lost the education and programming edge. This is because you grow better programmers by limiting their compute resources, not by throwing billions of borrowed dollars at them. And we are doing both — forcing competitors to be more resourceful, while focusing on near-term, not long-range goals.
I’ve been building my own local AI agents. It’s fun, it’s not easy, and the AI model truly matters. We’ve hit an inflection point where a good, detailed prompt is not enough — and in some cases may even limit your results. Coordinated agents are the current new thing. They are computationally expensive, but they leverage today’s AI capabilities better.
We’ll get back to a point of domain expertise driving the bus as use of custom agents grows. At least until we end up effectively uploading our brains (or at least our domain knowledge) into these systems, in the pursuit of income and success. Heaven help us.