Google I/O, the company’s developer conference, started today, but for the first time since I can remember, Android and Chrome OS have been relegated to day two of the conference. The first day was all about “AI”, most of which I’m not even remotely interested in, except of course where it related to Google’s operating system offerings.
And the company did have a few things to say about “AI” on Android, and the general gist is that yeah, they’re going to be stuffing it into every corner of the operating system. Google’s “AI” tool Gemini will be integrated deeply into Android, and you’ll be able to call up an overlay wherever you are in the operating system, and do things like summarise a PDF that’s on screen, summarise a YouTube video, generate images on the fly and drop them into emails and conversations, and so on.
A more interesting and helpful “AI” addition is using it to improve TalkBack, so that people with impaired vision can let the device describe images on the screen for them. Google claims TalkBack users come across about 90 images without description every day (!), so this is a massive improvement for people with impaired vision, and a genuinely helpful and worthwhile “AI” feature.
Creepier is that Google’s “AI” will also be able to listen along with your phone calls, and warn you if an ongoing conversation is a scamming attempt. If the person on the other end of the line claiming to be your bank asks you to move a bunch of money around to keep it safe, Gemini will pop up and warn you it’s a scam, since banks don’t ask you such things. Clever, sure, but also absolutely terrifying and definitely not something I’ll be turning on.
Google claims all of these features take place on-device, so privacy should be respected, but I’m always a bit unsure about such things staying that way in the future. Regardless, “AI” is coming to Android in a big way, but I’m just here wondering how much of it I’ll be able to turn off.
> Google claims all of these features take place on-device, so privacy should be respected, but I’m always a bit unsure about such things staying that way in the future.
Which is, of course, besides the point. Like microsoft, google is going to be doing the computing for their next level of “big data” abuse all on-device.
It is, of course, a complete coincidence that all that fancy “tensor” hardware they’ve been chucking into their devices also happens to be eminently suitable for this.
Please for the love of all that is holy, define the term AI, as in the real one.
Most of these “features” seem to be called extensions in Gemini settings, if you turn off Activity history for Gemini then all extensions are disabled, so that seems to be a quick way to disable it.