alt Hacker NewsHonestly I think their approach is the long term most useful one. Being SOTA is probably very costly and difficult. Optimising all the smaller stuff that's real world useful for people seems like it would have better long term use.
On top of that I would imagine the research side of Google might disk over things from one modality being useful to another - something like an audio optimising or memory optimising for text to speech could maybe also be useful for translation or world models. Etc etc
I can see where you are coming from. Vendor lock-in.
Basically spread many AI's everywhere, get people develop/user their ecosystem, and lock them in eventually.
But I find their models' intelligence lacking still.
I don't ever use their built-in gemini features (i have paid gmail) because they don't work well.
e.g. I ask gemini to format my Google docs per Google's material design spec with spacing, etc. It does a real bad job. Many times it does it line by line, and when I finaly get it do it for the whole doc, it does it sloppy, and extremely slow (takes 5 minutes for 10 page doc)