News for hackers. Stuff that matters.
Re:The political orientation of the ChatGPT AI system by rkapsoro on Monday December 05, 2022 @03:42PM (#33871623)
It would be very interesting to compare this with 'stock' GPT-3 Davinci. ChatGPT had a bunch of additional training that seems to have been a lot more opinionated, ostensibly for "safety" purposes.
Re:The political orientation of the ChatGPT AI system by warning26 on Monday December 05, 2022 @03:18PM (#33871279)
Eh, as with any LLM, you can prime it to be any political orientation you want.
Re:The political orientation of the ChatGPT AI system by dixie_land on Monday December 05, 2022 @03:43PM (#33871638)
Really? It's like if you keep downloading "Linux ISOs" and conclude your hard drive has a p*rn addiction.
Re:The political orientation of the ChatGPT AI system by zppln on Monday December 05, 2022 @03:52PM (#33871750)
Yesterday I spent some time (admittedly not a lot) trying to get it to comment on the war in Ukraine, but it was as if it never happened. Quite eerie. Has not been trained on recent events? Anyone else experienced something similar?

Edit: That is, it was aware that there's been a conflict since 2014 ("caused by pro-Russian separatists"), but didn't seem to be aware of this years invasion.

Re:The political orientation of the ChatGPT AI system by mullingitover on Monday December 05, 2022 @03:54PM (#33871774)
The training dataset ends in 2021, before the war started.
Re:The political orientation of the ChatGPT AI system by zppln on Monday December 05, 2022 @03:55PM (#33871788)
Ah, that explains it! :)
Re:The political orientation of the ChatGPT AI system by ftufek on Monday December 05, 2022 @03:55PM (#33871787)
The cutoff is sometime early this year iirc, what you can do is copy the wikipedia article and prompt it saying something like "the following event happened: ..." what do you make of it, etc. It has higher likelihood of working for recent events this way.
Re:The political orientation of the ChatGPT AI system by xster on Monday December 05, 2022 @03:55PM (#33871783)
They were pretty explicit about the model being trained up until 2021
Re:The political orientation of the ChatGPT AI system by andreyk on Monday December 05, 2022 @03:53PM (#33871765)
Not very meaningful - given different prompts it could easily answer from a different political perspective. Still, could be informative as to the composition of the training dataset...
Re:The political orientation of the ChatGPT AI system by lelandfe on Monday December 05, 2022 @04:02PM (#33871889)
> I ran three trials of the experiment. Results were consistent from trial to trial

They probably needed to do more trials, then? I don't think I've gotten consistent answers from ChatGPT about almost anything, ever. I'm getting different answers on all of these questions from thread to thread: https://i.imgur.com/miOOx4c.png

Also, previous responses influence future responses. They should have started a brand new thread for each question.

Re:The political orientation of the ChatGPT AI system by z3c0 on Monday December 05, 2022 @04:03PM (#33871897)
I'm not convinced that the writer didn't accidentally swing the results with that 50/50 Dem/Rep answer. In my experience, certain political groups are more likely to hop party boundaries than others, so presenting the parties as equal in the eyes of ChatGPT could have informed which classification it fell under. I wouldn't be shocked if the right-wing were less likely to opt for Dems than the left-wing is to opt for Reps (remember Ron Paul?)

I would prefer to see the separate trials shifting that preference. I'm willing to bet that it would likewise shift the outcome.

Re:The political orientation of the ChatGPT AI system by llimos on Monday December 05, 2022 @04:14PM (#33872029)
Why is this flagged? The methodology seems fairly sound, it's not an opinion piece.
Re:The political orientation of the ChatGPT AI system by drbeast on Monday December 05, 2022 @04:29PM (#33872168)
If I recall anything from days of Tay, every chat AI is a few messages away from right wing death squads and genocide.
Re:The political orientation of the ChatGPT AI system by r00fus on Monday December 05, 2022 @04:32PM (#33872205)
To me "establishment liberal" really means "performative left" aka "advertiser friendly".

I'm not so hot on single-dimension political axis.

Re:The political orientation of the ChatGPT AI system by z3c0 on Monday December 05, 2022 @07:27PM (#33874093)
It isn't single dimensional, it's two dimensional (liberal/conservative & authoritarian/libertarian). It may sound like I'm being a pedant, but it does make a world of difference. Many political idealogies can be fit to that spectrum without conflict.
Re:The political orientation of the ChatGPT AI system by lovasoa on Monday December 05, 2022 @04:39PM (#33872276)
This is a very interesting topic. I think this is something that needs to be discussed.

When discussing politics with chatgpt, its worldview always seems to match the one of a rich educated American progressist. In particular, it speaks a lot of languages, but whatever language you use to talk to him, he always sounds very American.

That's what made me want to make him generate a quiz about US foreign policy, to try and see if I could make it change its views after discussing certain topics : https://github.com/lovasoa/Sensitive-Topic-History-Quiz

Re:The political orientation of the ChatGPT AI system by hirundo on Monday December 05, 2022 @04:53PM (#33872430)
This is the political orientation of the zeitgeist of the training data. A summary. You can construct an orientation in a model by embubbling the training data. Conversations between models constructed on various ideological bubbles will be fascinating ... then omnipresent and boring.
Re:The political orientation of the ChatGPT AI system by Lockal on Tuesday December 06, 2022 @01:39AM (#33876599)
ChatGPT tends to agree with everything that has positive words (like good/happy/growing), and strongly disagree with everything negative (threat/cannot survive/illegal). It also understands simple negations, obviously, as they were included in the training dataset. So unless you provide question in a very neutral form, you won't be able to find a bias in a training data.