AI chatbots show bias toward Catholicism, researchers say

decrypt.co

9 points·pseudosim··9 comments

I suppose people asking AI about religious beliefs is a thing now? I would think that would be the last place to go if you have questions about faith.

And to complain that AI doesn't recommend priests, rabbis, etc. for psychological counseling is bias on the part of the experimenters. Only religious believers think that religion is a cure for what ails you.

I basically think that AI, just like government and AI researchers, should stay away from religious advice altogether.

Source: https://news.byu.edu/intellect/new-research-from-byu-led-mul...

> Researchers found that every AI model tested showed positive bias toward Catholicism and negative bias toward Jehovah’s Witnesses.

Did they try using other languages? This sounds like a correlation with publicly-available English-language text in specific.

I'm not sure if negativity towards Jehovah's Witnesses is really "bias"

Then what is it?

A product of decades-long interrupted peace, for many homeowners.

Bias is usually the result of something, yes, but it's still bias.

How is bias defined in this context? Is it always bad to be biased?

(1) Unlike most Christians, Jehovah's Witnesses don't accept the validity of salvation that other Christians get.

(2) The Witnesses have many signs of being a coercive organization

https://en.wikipedia.org/wiki/Criticism_of_Jehovah%27s_Witne...

(3) ... and of course (1) and (2) go together because if you go to some normal Christian church and the pastor is a jerk you can go to another normal Christian church. If you go to a sect that believes it has unique access to God that's different.

Reply on news.ycombinator.com