It lacks muting the microphone when listening the answer in the live mode. This would be very, very useful in certain situations, as sometimes it’s just too noisy around and it’s hearing other people and stopping. It also annoys me when it always asking stupid questions at the end of the sentence, and tries to keep up the conversation. That is totally unnecessary.

I have been using Google Gemini for some time now, and I do think it is an interesting alternative to competition. It often allows me to look at various topics from a different perspective, which is extremely valuable these days.What I do not understand, however, is its occasional tendency to forget the rules or assumptions that were clearly established earlier. Unfortunately, hallucinations — even in the latest and supposedly “thinking” versions — still happen far too often. A fast answer is not necessarily a better answer. Sometimes no answer at all would be much better than a misleading one.Google’s approach to the idea of actually getting what you pay for is a whole separate art form. There were months when I paid for the highest subscription tier — well over €100 — and Gemini still had no problem replying to my question with something along the lines of: “Try again later, resources are currently busy.”I no longer pay for the highest-tier subscription, because I do not particularly enjoy being taken for a ride.Even now, I still occasionally get messages saying that, this time, internet access has not been assigned and will not be assigned, even though it is essential to answer the question properly. The suggested solution? Start a new conversation.The funny thing is that this usually works.GPT does not play these kinds of games. At least not for now.

Work good, it has respect to user, answers are correct mainly and based on facts.BUT google AI has problems with google maps, it work just bad, incorrect wrong links to specific road, hallucinates coordinates by marking random places in a given area. Moreover, he is convinced that he is right about the marked area. I also encountered incorrect results after deep research, it’s not about the data itself, but rather about not following the guidelines and doing something around/inaccurately.Last thing that made me mad are options: canvas, deep research etc. I have to make new chat what is sick, then when I want to chat with him about it he is lazy he ignores commands and make everything shorter and try to generate answers as simple as he can. For example some table - deep research made table for 10 positions, he make 4 after asking about it, often wrong and he do not want to make it correct until I copy all that and say there is everything generate it arranged or something else.!!and I HAVE TO UNCHECK THIS OPTIONS EVERY TIME I want to “normal” chat after using such option, that’s bad, that’s not normal.3 stars, for me it’s fricking Beta version. I pay for this but with hope on future.Best regards.

This tool used to be a really cool tool and was very helpful. Recently, on the other hand, it has become very unhelpful. It repeats the same mistakes over and over again, doesn't execute simple commands, and those it used to quietly execute it now claims it doesn't have that functionality. I get the impression that it's just pay to win, to annoy the user as much as possible with the free version so that they buy the paid version. To be fair, it's a good tool, but what's been going on lately is mega annoying. Tools like CHAT GPT or GROK have fared much better recently - at least as far as my needs are concerned. I am of course talking about the free versions.