For me, one of the big problems with AI is that it doesn't run in a sandbox on your computer. It all happens far away, in a server farm that might be in the US, or India, or Idon'tCareAboutYourPrivateRightsIstan. The data travels, usually unencrypted, over the interwebs, then who knows what happens to it in FarAwayIstan.
The other big problem is that the systems are self-programming, and you have no idea what logic or algorithms are being applied to your data.
And what happens when it gets your data mixed up with your neighbour's novel about blowing up the gummint? Especially given AI's well known inability to tell the difference between truth and lies, between fact and fiction.
When I am searching for something, I am only interested in its file name. I don't want some software looking inside all of my documents, and analysing my photos, then applying who knows what algorithms to them.
There can be no failsafes with AI, as nobody knows what is going on inside the box.
That's why, for the time being, I am sticking with my Raspberry Pi and using a well-known version of Linux.
Retro me, Windows!!!