| My impression is they haven’t focused much on general purpose AI. But Apple actually has a lot of very good AI models sprinkled throughout its products. Just a few that come to mind… - Tap and hold an object in Photos and it will figure out how to separate it from the background for you - AirPods noise cancelation - iOS 17 autocorrect is based on a transformer model and works noticeably better - Optimized Battery Charging, which learns your charging habits and tries to delay putting a full charge into your battery until just before you unplug, in order to avoid damaging the battery - Detection Mode is an awesome accessibility feature where you point the camera at something and it will describe what it sees Apple calls all of these things Machine Learning instead of AI and they are all optional features within an existing product. Seems like a very deliberate strategy. But they are utilizing the latest techniques. CoreML and the M series chips are also very competent at training and using AI models. Maybe the reason Siri is stuck in the dark ages is because it would be entirely AI dependent. They could have a “no generative AI” mode but nobody would use it. I’m guessing Apple is looking for a breakthrough in how to prevent it from hallucinating / lying. |