Google’s Gemini mess!

Google can’t stop messing up

Google’s wobbly journey to introduce any useful AI product started with their fake Gemini demo a few months back. Claiming features that do not exist:

https://techcrunch.com/2023/12/07/googles-best-gemini-demo-was-faked/

Then questionable quality:

And now this:

X (aka Twitter) is filled with created images that are either down right wrong or, in some cases very questionable. I can’t paste examples of the questionable images Gemini has been creating. But it’s incredibly easy for anyone to search for them. Here are two embarrassing ones:

What’s really offensive and questionable is Jacks’ response:

Historical contexts have more nuance to them and we will further tune to accommodate that. This is part of the alignment process — iteration on feedback

Not only he does not apologize for the offensive content Gemini has been creating, he claims that this is just a “nuance” and part of “iteration process”. The fact is that Google has decided to create so much bias in the model that it’s just offensive. Not owning up to their mess is yet another evidence Google is in deep trouble.