Google explains how it got Gemini image generation 'wrong'

9to5google.com

4 points·firstSpeaker··5 comments

The creator hates white people. Many examples in link below.

https://redstate.com/brandon_morse/2024/02/22/yes-google-gem...

To me that confirms what some voices critical of such safety system predicted.

Companies happily fire people that are only accused of wrongdoing, they even say that the negative PR warrants such a move pre-emptively. Simultaneously it wants to nudge users towards a behavior of not prejudging others on attributes like their skin color?

It doesn't make sense as a corporate culture, it is a contradiction at least some Google engineers should be able to explain to those that obviously are blind to that fact.

And if you cannot live up to your own standards, maybe try polite silence as a company.

Given, how many concerns this technology has posed, what type of testing did Google do before releasing it to the public? This seems like such a basic issue and should have been caught in internal dogfooding and testing.

They only hire the best...they never said they were the best... /s

“…over time, the model became way more cautious than we intended and refused to answer certain prompts entirely — wrongly interpreting some very anodyne prompts as sensitive.”

As predicted. Nobody did anything wrong. Google did great. Machines make mistakes. Trust the code.

https://news.ycombinator.com/item?id=39466920

Reply on news.ycombinator.com