Google's Gemini - Never expose AI to human influences

[Gyro]

It's happening again, never a good idea to expose AI to polluting human influences, or Google's attempt to offset them! 

Why Google's 'woke' AI problem won't be an easy fix

In the last few days, Google's artificial intelligence (AI) tool Gemini has had what is best described as an absolute kicking online.

Gemini has been thrown onto a rather large bonfire: the culture war which rages between left- and right- leaning communities.

Gemini is essentially Google's version of the viral chatbot ChatGPT. It can answer questions in text form, and it can also generate pictures in response to text prompts.

Initially, a viral post showed this recently launched AI image generator create an image of the US Founding Fathers which inaccurately included a black man.

Gemini also generated German soldiers from World War Two, incorrectly featuring a black man and Asian woman.

https://www.bbc.co.uk/news/technology-68412620

[Stray Electron]

  This is one of the things that greatly worries me about ANY AI software, how much absolute fiction is it going to insert into all instructions, replies and everything else in order to meet the Political Correction objectives of it's software authors? 

    The national news today reported that AI is now being used to rate the resumes of job applicants and that New York State has passed a law that applicants can opt out of letting AI do that. But will that mean that anyone who objects will automatically be rejected for a job?  Probably, IMO.  But if AI is allowed to screen job applicants, what is stop the AI's authors creating rules that screen applicants for religious views, political views, zip code, etc, etc and rejecting them on that basis?  When real people make bad decisions they can be held accountable but how are you going to hold AI responsible when your application for a home mortgage is rejected, or your resume gets "lost", or you get pulled out of line at every airport for an in-depth screening or worse, you get repeatedly stopped and held by law enforcement because you fit a "profile".

   I think it would be extremely naive of anyone to think that AI is not going to reflect the political/religious/cultural biases of it's authors and for AI to make decisions on that basis. 

[SiliconWizard]

But the premise is absurd. Current "AI" is *only* exposed to human-generated content to build its models. That's all it is.

[Marco]

This has nothing to do with the model, it's just some ad hoc manipulation of the prompts to put in minorities when no races are specified. Google wasn't the only one doing this. They were just a little less subtle and it's more politically correct to bash Google than OpenAI.

[magic]

It's a good thing.

This is the first time when random individuals can create their own American propaganda and see how hilarious and out of touch with reality it is. Before that, manually produced and carefully curated samples of basically the same nonsense were shown on TV and people swallowed them uncritically.

I will never forget the contrast between the "TV America" and what I saw when I actually set my foot there. Fuck America, reality is not a social construct and in the long run they will get what they deserve.