I was trying to use Copilot today in an AVD session and it gave me repeated errors when I was typing in the query box. Ultimately I could not use it due to the errors. Relaunching the app, changing to a different type of AVD session didn’t fix it. It is still broken.
Now if I was a paying enterprise customer having this problem I would be angry. I am not getting what I paid for, and I couldn’t even access past content of past chats. AI has honestly cost me more time and energy than it has saved me.
Instead of calling it AI they should call it, random stuff might happen and feel lucky is partly works at all. This is a disappointing feature from Microsoft, and the ChatGPT 5 isn’t much better.
I used ChatGPT5 extensively on Saturday to test it and while it asked if I wanted to do a deeper dive, when I asked it to do that, it didn’t find anything useful. It should have pre-fetched content and then said there is nothing to find or something to find.
I also was irritated that ChatGPT 5 equivocated in its answers. Just so you know it says its does a fresh search with each query. Yes it does, but it has no problems misunderstanding old content and hallucinating things that aren’t true. When I challenged it, it admitted that it should have read it more closely.
If users lose confidence that the tool is reliable, then your company is done. If we have to double check an answer, it really makes it almost useless. At least to me AI is becoming less useful than it used to be.