1,000 AIs in your pocket

When you’re late to the game, do less, better. Apple used to know this. When the original iPod arrived at my house, I could tell it was a breakthrough. Famously less space than a Nomad, but an innovative UI, great design, and fast FireWire. It was so good it set in motion everything else for the company’s current success.

It’s easy to look back now and judge Apple’s AI rollout, but even at WWDC you could tell Apple was throwing everything at the wall. Image Playground wasn’t going to be as good as frontier image models. Siri world knowledge wasn’t going to be as deep as what ChatGPT could do. By trying to do nearly everything, each piece feels like a gimmick.

There are some useful features in Apple Intelligence. Even flawed, I like notification summaries. But the good is getting lost in the noise. It seems clear now that Apple should’ve taken the publicity hit last year for not yet having an answer for every generative AI capability. They should’ve resisted scrambling to do too much, instead focusing only on what their models could knock out of the park.

Manton Reece @manton