And I hope the reverse. Most of my data I am happy to entrust to Apple, and I don't want them to release substandard not-very-smart products just because they're limited to AI models that only take a few gigs of RAM and can run on a single phone processor.
Please just use some huge models to make actually smart products and run them in the cloud if it isn't feasible to run them on-device. Perhaps have an 'offline mode', which runs small models (the google assistant already does that - and it's very noticeable that the online mode has very accurate speech recognition, whereas offline mode can only recognise basic words reliably).
Please just use some huge models to make actually smart products and run them in the cloud if it isn't feasible to run them on-device. Perhaps have an 'offline mode', which runs small models (the google assistant already does that - and it's very noticeable that the online mode has very accurate speech recognition, whereas offline mode can only recognise basic words reliably).