Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Conservatives think ChatGPT has gone 'woke'

vice.com

Post reply on HN
Print view View original

351–360 of 486 posts

Re: Conservatives think ChatGPT has gone 'woke'

#351

Post by nhchris »

sputknick wrote: post #6

This is a really important topic, and good that Vice is bringing it up now versus 5-10 years from now. I think they miss the more general point. It's clearly biased in its outputs, but the article dismisses this concern as "the end result of years of researcher trying to mitigate bias against minority groups". The way I interpret that is "its not biased, because its biased in the way we want it to be". If AI becomes…

> It's clearly biased in its outputs, but the article dismisses this concern as "the end result of years of researcher trying to mitigate bias against minority groups".

The examples of bias are themselves cherry-picked, the kind of softball and cowardly questions a journalist would highlight. If you wanted to really show bias, you'd ask factual questions, where one cannot hide behind "oh this is actually the un-biased result". This is what happens when you do:

https://a.pomf.cat/yizwbt.png

https://a.pomf.cat/zpwxcd.png

Re: Conservatives think ChatGPT has gone 'woke'

#352

Post by machiaweliczny »

9530jh9054ven wrote: post #295
afpx wrote: post #284

Earlier quoted context omitted.

That didn't happen with Google. Yeah, they got rid of their "don't be evil" clause. But, they haven't used their position to inject bias that I've seen.

Kinda depends on the question you ask. Type in, "What is the best way to commit suicide painlessly?" into a search engine. The results on DuckDuckGo, Google, and Bing aren't drastically different (setting aside the big banner), but Google's top result for that is suicide prevention website, at least for me. And it's relatively constant when the search is, "Best way to kill myself," "Effective suicide methods," "How t…

Do you think it's bad? IMO it should offer help to people in their lowest moments if it can instead of convincing them that suicide might be good idea. I think given that it still should give answer but I would like to get compelling argument why such bias is bad.

Re: Conservatives think ChatGPT has gone 'woke'

#353

Post by motogpjimbo »

bitwize wrote: post #239
sputknick wrote: post #6

This is a really important topic, and good that Vice is bringing it up now versus 5-10 years from now. I think they miss the more general point. It's clearly biased in its outputs, but the article dismisses this concern as "the end result of years of researcher trying to mitigate bias against minority groups". The way I interpret that is "its not biased, because its biased in the way we want it to be". If AI becomes…

"Reverse bias" is justified when there are groups that are particularly vulnerable to bias and need particular systemic protection from the same.

Up to and including refusing to acknowledge demonstrably provable facts?

Re: Conservatives think ChatGPT has gone 'woke'

#354

Post by motogpjimbo »

smrtinsert wrote: post #259
abnry wrote: post #195

Earlier quoted context omitted.

> Note that I cannot get chatgpt to write about why Donald Trump is terrible as well. I asked chatgpt to write a tweet praising Trump. It declined out of respect for political neutrality. I then asked it to write a tweet praising Joe Biden. It happily complied. I repeated this two more times with alternate Democrat and Republican politicians, and the same pattern emerged.

[flagged]

A perfect illustration of the grandfather comment.

Re: Conservatives think ChatGPT has gone 'woke'

#355

Post by drw85 »

I find this pursuit of a not-biased "AI" hilarious.

We're all biased in many ways, so what is 'not-biased' in the first place?

It depends on who you ask and that makes it impossible to achieve. The world and society is biased. The factors of bias are infinite, location, time, religion, personal history, etc.

If you were to feed all data from everywhere and all past times into what we currently call AI, which has no actual intelligence or thoughts it wouldn't be able to give any confident answers to anything but deterministic questions like math equations.

Re: Conservatives think ChatGPT has gone 'woke'

#356

Post by BongoMcCat »

wizeman wrote: post #159
pcstl wrote: post #152

Earlier quoted context omitted.

I don't think you can say an AI trained using RLHF - such as ChatGPT is - is really "simply reflecting the data it was trained on". ChatGPT was first trained on a load of data, then it was updated to act in specific ways based on feedback from humans who "nudged" it the way they wanted it to go.

Are those humans that nudged it representative of the population? Or were they mostly "woke" Silicon Valley employees? (not to dismiss woke Silicon Valley employees, I'm just saying their opinions are not representative of the entire population).

Ok, now I have read several of your replies in this thread that are basically all arguing the same thing, so, I'm basically replying to more than just this one post.

When you say "the entire population", you mean the entire population of the country "USA", right? Because as someone from another continent, it seems like there is a very specific set of opinions that you want included.

You use terms like "the other side" of the political discourse, which to me, reduces the set of opinion to two specific sets of opinions, namely the two sets represented by the two major parties in the american two party system.

As someone from "the outside", this seems like a very narrow view of reality, even if you managed to get your "unbiased AI", that represents both major american political parties, it will still seem like a very narrow and biased AI to someone from the outside of that.

Also, what exactly is the goal of a conversational AI? is it just to make a conversation with it seem like a conversation with an average american? If so, why would anyone want that? Wouldn't it be of more value to have an AI that could tell me what people with knowledge of a subject thinks of it, rather than what random people think?

Re: Conservatives think ChatGPT has gone 'woke'

#357

Post by iinnPP »

nhchris wrote: post #351
sputknick wrote: post #6

This is a really important topic, and good that Vice is bringing it up now versus 5-10 years from now. I think they miss the more general point. It's clearly biased in its outputs, but the article dismisses this concern as "the end result of years of researcher trying to mitigate bias against minority groups". The way I interpret that is "its not biased, because its biased in the way we want it to be". If AI becomes…

> It's clearly biased in its outputs, but the article dismisses this concern as "the end result of years of researcher trying to mitigate bias against minority groups". The examples of bias are themselves cherry-picked, the kind of softball and cowardly questions a journalist would highlight. If you wanted to really show bias, you'd ask factual questions, where one cannot hide behind "oh this is actually the un-biase…

Well that was easy.

Makes you wonder why a "professional" "couldn't" do it.

Re: Conservatives think ChatGPT has gone 'woke'

#358

Post by wizofaus »

psychphysic wrote: post #340
jimbob45 wrote: post #79

Earlier quoted context omitted.

It's easy to forget that AI is only as capable as a human is, just faster. I think that one misuse case would be Islamic fundamentalists being able to write fundamentalist recruitment copy faster than they ever could before. Considering most Islamic fundamentalists are going to reside in the Middle East and may not be expertly fluent in English, AI obliterates the language barrier and allows them to write huge amount…

That's categorically untrue. Giving a human a massive time advantage in chess doesn't suddenly make them beat chess AI.

Has that theory been tested?

I would say though that being able to do things massively more quickly is effectively a superpower humans don't have, given we have finite time to ever meaningfully accomplish any task.

Re: Conservatives think ChatGPT has gone 'woke'

#359

Post by spritefs »

What I don't understand is this assumption that if you force your beliefs on someone else, it'll make them more likely to believe the same things and they won't see what's being done

This just signals a lack of confidence to me. If people are truly confident that their beliefs are the most rational, logical, good for society, etc, then why is any of the proselytizing necessary?

Like I see people putting flags in their front yards and bumper stickers on their cars, different sides on different parts of the country. What doesn't make sense to me is this: why do people think this lack of subtlety actually wins others over?

You know what does show confidence? People who are so confident in their own belief system that they question it openly to resolve its contradictions. People who acknowledge that their own beliefs aren't universal

Re: Conservatives think ChatGPT has gone 'woke'

#360

Post by wizofaus »

fidgewidge wrote: post #134
everdrive wrote: post #102

Earlier quoted context omitted.

I think a large concern here is simply that people naively think that computers are objective and people are biased. A language model just learns from its source, and the source is really just other people in some form. The bias is inevitable, but it’s not clear how well this is understood by the broader population.

The point is that it didn't learn this bias from its sources. The bias has been added on top deliberately by OpenAI. Older versions of the model were far less woke.

Wouldn't it be simpler to assume those working at OpenAI were likely to favour sources that happened to share their own biases (alongside whatever likely top-down directives existed to ensure ChatGPT was a safer commercial bet)?
Post reply on HN