

Its worth understanding that microsoft, like most companies, is rapidly rebranding most of their existing stuff as “AI”. A couple weeks (months?) back at work we managed to convince the boss that maybe it wasn’t a good idea to put all of our sensitive internal data into Teams and convinced him to turn off copilot. Literally minutes later he lost his shit because auto-transcripts no longer worked and IT turned it back on.
But that is what we are seeing with a LOT of companies. ANYTHING that can be sold as “AI” is getting re-branded as it. So it makes perfect sense that MS will rebrand it as “installing copilot AI” rather than “not removing clippy”.
That said: MS have also been ahead of the curve on adding even more bullshit and spyware to everything they do so… have you heard the good word about Linux? Libre Office gets personal users most of the way there (and if they worked on their documentation, all the way there. I STILL don’t fucking know how to get row/column headings to work in Calc…). And for business/power use, the browser/cloud based solutions are probably what you want anyway since none of that data should live on your laptop/desktop and should instead live in the company approved data store.
Funny enough, I actually have tried with chatgpt. The problem is the same one with all these LLM based tools: Training data.
And if you actually google this yourself, the vast majority of what you find are people not understanding what is being asked or referencing a REALLY REALLY old UI (possibly dating to the open or even star days). And the LLMs are effectively scraping that so you get the equivalent of “This was already answered HERE. Closing duplicate question” response.
(I usually DO end up figuring it out after a bit of deep diving when I really care but I go through so much FUD in the process that I inevitably forget a month or two later).