I am trying to stay open-minded here.
As it is, I don’t see a usage that justify install a privacy nightmare on the system level, an app, or even a browser extension.
The only time I have been using any AI stuff lately is when Brave_Search/Copilot_on_Bing do a quick summary of the search results, and give me the link to the website. It don’t need to constantly see what I am doing for that.
I use it for troubleshooting weird errors, explaining my point when I can’t communicate well, connecting dots. For example, finding common points that link myths and legends of an unfamiliar culture, to get starting points on what to read about. This would otherwise be hours of research sifting through unrelated stuff just to find some mention of sparrows or whathaveyou.
I’m excited for AI as a technology, but not for this. I was expecting to have one trained to parse system logs and stuff, to help people troubleshoot on-device using natural language. This thing will take screenshots every few seconds? What if it misses an error screen because you close it too fast? Can it still function on a BSOD? What about 2k and 4k monitors, which will produce much larger screenshots, paired with lower-capacity SSDs for cost reasons?
I could go on, but this is just boneheaded on every level that I can see.