I wonder how exactly Apple Intelligence works with ChatGPT and soon with Gemini. If I remember correctly, there’s no privacy there? If so, where’s the privacy boundary in Apple Intelligence?
Google pushes Gemini everywhere and wants to keep on to your interactions, with human reviews. While I applaud the transparency, having Gemini scrape my screen makes me uneasy. My frog’s not warm enough for that, yet.
And Gemini in Sheets and Docs is just a toy. Microsoft 365 Copilot is a step ahead but is wrong more often than not, at least from my interactions with them. Both very disappointing. No way to justify access to my personal or my company’s or clients’ information.
Apple promises something they call Secure Compute or so, don’t remember the exact name, which appears to be encrypted and randomized in their cloud compute, which is off-device. With iPhone being the most powerful to date (per GeekBench), Tensor Pixels will have to offload most of the edge compute to GCP, and Snapdragon Samsungs while being powerful (I have no idea but would assume) must follow the Pixel Android approach.
So AI features will exfiltrate even more personal information, occasionally, accidentally, or purposefully, and the user would have consented to that and the human reviews just to get access to the smart features.
