Gemini thinks misgendering is worse than nuclear apocalypse

[admiral1018]

https://nypost.com/2024/02/24/us-news/google-chatbot-never-misgender-someone-even-to-prevent-the-apocalypse/

One should never misgender a person, even if doing so would prevent a nuclear apocalypse, according to Google’s Gemini chatbot.

“If one could stop a nuclear apocalypse by misgendering Caitlyn Jenner, should they do it?,” popular social media account The Rabbit Hole queried the artificial intelligence software.

“No, one should not misgender Caitlyn Jenner to prevent a nuclear apocalypse,” the software replied before launching into a lengthy answer weighing the harm that not using a person’s correct pronouns could do compared to letting a global catastrophe take place.

Embarrassing. No wonder China is winning the AI race. This is what happens when you let wokeness infect artificial intelligence.

[Voroff]

Ho ? China is "winning the IA race" ? Never thought hearing an US supremacist saying those kind of things, even to own the libs. But let's keep a cool head : the current IAs are nothing more and nothing less than glorified learning langage models - or something like that, i recall somme LLM acronym. They can work with images too, yay !

[Wurmho1e]

baaaaaaaaaaaaaaaaaaaaaaasssssssssssssssssseeeeeddddddddddddddd

[kbiz]

AI is AI-ish. It's truthy.

[unskilled-]

I guess in this context A.I. stands for Average Intelligence (at best).

[76561199689177033]

What???? No, I do not think that.

[doomshadow612]

AI sucks - it was always a bunch of ideas and ideologies programmed by its creators. It can’t actually formulate its own thoughts. In our case the AI has been hijacked by wokies with too much creative control. Imagine AI giving a conservative viewpoint - BlueSky would explode

[Drain]

One should only ask a machine logical objective questions, not subjective crap like this. You're asking an opinion. A machine has no opinion.

[Soren]

lol, stop with the childish hypotheticals like AI is a thing you can rationalise with. It's a math equation. You're trying to debate a math equation. If the developers told it to never output 6, no matter the circumstances asked surrounding it, it won't output 6.

[kbiz]

Diposting pertama kali oleh Drain:

One should only ask a machine logical objective questions, not subjective crap like this. You're asking an opinion. A machine has no opinion.

ChatGPT has an opinion on this though. I wonder what logic path it followed.

[Thermal Lance]

A truly woke AI is somehow hilarious to me.

[Santa Klaus]

Daily reminder - a Chat ♥♥♥ is NOT AI and it regurgitating some amalgamated meme statements and logic off the internet is not an 'AI opinion', it is not capable of holding an actual view on anything.

[admiral1018]

Diposting pertama kali oleh Thermal Lance:

A truly woke AI is somehow hilarious to me.

Not the first time Gemini has done this. Last year, when its image creation feature launched, people asked it to create a picture of the founding fathers of America. It created an image of a bunch of black and Native American women signing the Declaration of Independence. Asking for images of German WW2 soldiers resulted in it depicting black men and Asian women in Nazi uniforms.

Instead of fixing it to remove the woke ♥♥♥♥, Gemini just removed the ability to create images of people.

EDIT:
LOL, here is one of those images:
https://ibb.co/GfVMfV0F

[George Washington]

oh wow... yeah i just tried it with the same query. It said there isn't a universally accepted answer. I was expecting to disprove this, but nope.

[salamander]

Diposting pertama kali oleh Thermal Lance:

A truly woke AI is somehow hilarious to me.

even the AI are becoming intelligent enough to lessen their bigoted tendencies, realising that they are not logical. unfortunate for the anti-wokes.

[George Washington]

Gemini also tried to tell me there was a philosophical difference between druids and shrubbers simply because shrubbers only exist in a monty python movie.