New optimization makes GoogleAI feel lobotomized

[ripe mirage]

I understand that this is a basic need, but all users feel that Gemini 3.0 and Nano Banano Pro got nerfed..
The responses from launch were really good and better than the competitors. And right now it became even worse than 2.5 Flash Lite... It basically do loses for Google because of the bad intelligence: Users will not continue their "trial" subscriptions. And it's hallucinating a lot, which wasn't even a thing at launch of 3.0 Pro. Now I'm asking it questions and it answers in most cases wrong, but at launch it really made good answers. 🍌 didn't loose the quality too much, but it lost some abilities in understanding the prompt and it barely follows it and I need to do many tries to do what I want..

It's not hate to your side.. I actually like gemini products and want to make them better.

(I don't think that I'm being paranoiac since many users feel that, not only me, just check subreddits of your ai products)

[near crypt]

Hi there @ripe mirage ! We value your feedback, and we're always looking to improve our tools. If you could share some examples of the issues you've been facing with Gemini and Nano Banana recently (videos, screenshots, etc.), I'd be glad to forward them to the right people!

[ripe mirage]

This is just a slight mistake, but I had more worse cases, which I forgot actually where, sorry. This is latest that I found, I have too many chats

[ripe mirage]

Which I cannot share for some reasons

[ripe mirage]

Some case includes irrelevant info, but I asked gemini to google and it didn't

Or it did but didn't really include that

Maybe it's more about following the prompt

And in code it often deletes parts which I didn't ask to delete or even includes too many errors

It's also not doing coding properly and missing some implementations

[pine abyss]

Gemini accesses the actual google databases, it doesn't run the sort of google search you do (which also is just accessing the google database but in a more limited way than gemini can)

[ripe mirage]

Then gemini should be better that search right?

[pine abyss]

not necessarily - and for the record, when you talk to gemini in google search, it works better as gemini can see what you just searched on and is already focused on that topic and those tokens

[ripe mirage]

Google search ai is worser than gemini 3 pro

It can't even define truth

[pine abyss]

okay, well, that might be because you're treating the AI as if it's a data base search engine and not a thinking, intellegent entity

[ripe mirage]

I meant that google search ai uses smaller version to response faster

also gemini 3.0 pro have reasoning and google search ai not

[pine abyss]

that has very little to do with anything