Steve Troughton-SmithApple's Neural Engine evidently doesn't count for a whole lot if even basic on-device LLM text generation only works on an iPhone 15 Pro or M-series iPad? We're still in a situation where any developer who wants to actually make use of AI in their apps is going to need to integrate OpenAI's web API anyway, sending all your requests to the cloud, so all of Apple's privacy talk means nothing 🫠 Why even launch like this? I am baffled by the WWDC announcements
Wayne Robinson@stroughtonsmith I think there's three tiers. On-device, Apple Cloud & OpenAI. I think the only things being sent to OpenAI are the explicitly ChatGPT features with an explicit confirmation.
@iamwcr the feature, in total, only works on these devices. You misread my post
@stroughtonsmith ahh, sorry
metafizikal@stroughtonsmith @iamwcr could simply be a RAM limitation?
Denis@stroughtonsmith @iamwcr Reading the footnote to that graphic on Apple’s website it is possible that only the beta is restricted to these devices?
“Some features, additional languages, and *platforms* will be coming over the course of the next year.”
Zabby@denis @stroughtonsmith @iamwcr I doubt it. Would certainly be wild to eventually release those features on older devices and not had months of beta testers. By platforms I believe they mean the use of other AI companies, Google’s Gemini for example.
@zabby_troon @stroughtonsmith @iamwcr I doubt it as well, but hope springs eternal. I think platforms could also mean the other Apple platforms i.e. Apple Watch or Apple TV that were initially excluded from AI.