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

Political bias in AI: Where the AI models stand

trakkr.ai

Post reply on HN
Print view View original

241–250 of 346 posts

Re: Political bias in AI: Where the AI models stand

#241

Post by kurthr »

DoctorOW wrote: post #234

All this debate about how politics are defined or how bias is measured. My question is why should I care? Grok identifying as Mechahitler didn't drag me any further right, why would ChatGPT be able to drag me left? If you vote the way AI tells you, that's already a problem regardless of what it said.

I sort of agree with this in the abstract. The problem is that concretely, these LLMs are being used to decide whether you receive healthcare, government benefits, or whether your job/agency gets cut. So they have already had real world consequences due to DOGE, Insurance companies, and other uses. I certainly don't find either the methodology or graphical conclusions of this link very valuable.

One can argue what fraction of the several million children's (under 5) deaths are (going to be) due to cuts by DOGE to USAID (and later congressional appropriations) from Grok recommendations/justifications, what fraction were politically pre-determined, and which are "just deserts", but it would be hard to put it at zero.

https://ph.ucla.edu/news-events/news/research-finds-more-14-...

Re: Political bias in AI: Where the AI models stand

#242

Post by BLKNSLVR »

crooked-v wrote: post #204
yodsanklai wrote: post #194

Earlier quoted context omitted.

> can have views on all four quadrants but you'd never know that if I end up in any of all four Curious of examples of views falling on all four quadrants (not close to the center) without too much cognitive dissonance.

The problem is the "without too much cognitive dissonance" part. Lots of voters have completely dissonant views that they've never consciously considered in any way. Just by posting on HN, the people here are probably in the top 5% of conscious political awareness in the US.

That's a good point.

The secondary (and on from there) effects of most of their views are not considered, and so their views are not internally consistent at all.

Small government! (why do I have to wait so long for my support payment query to be answered?)

No immigration! (why can't I find someone to clean my house on the cheap?)

Re: Political bias in AI: Where the AI models stand

#243

Post by OkayPhysicist »

samat wrote: post #22

Real politics is 1% versus everyone. Mortgage crisis, financial bailout, inflation, taxing of labor and not the assets and assets capture by tiny percent of the population — see what MSM is pushing. This left vs right divide might been useful decades ago, but today is absolutely divide and control tactics

Left vs. Right largely boils down to "are hierarchies good?". A quintessential far-right belief is the divine right of kings. The modern incarnation of that is that "The 1% earned their power and influence".

Now, there are far right people who want to dethrone the current powers that be. But distinguishing between "I think Mark Zuckerberg shouldn't be given unlimited power because no one should have that much power over others" (left) versus "I think Mark Zuckerberg shouldn't be given unlimited power because he's a Jew" (right) is a useful distinction. One wants to eliminate the power structure, the other just wants different people in it.

Re: Political bias in AI: Where the AI models stand

#244

Post by cramer4next »

17761989 wrote: post #226
cramer4next wrote: post #175

Earlier quoted context omitted.

I tend to disagree. What it really comes down to is how you create your prompt. For instance whenever I'm looking left wing extreamisum I provide it with a few cases first. Like BLM riots in Minnesota, paid agitators, mask wearing during peaceful protests, etc. It then gives you a clear unbiased deprogrammed view of the reality on all the AI models.

[dead]

How could a prompt create a bias in a model that implements AI bias safeguards that are the systemic, technical, and governance controls used to prevent algorithms from producing discriminatory, unfair, or stereotypical outcomes.

I've never in the 3+ years I've been using AI been able to craft a prompt to influence a bias output.

Its all about the ingested dataset and guardrails. The "prompt" is meant to extract data, not change it.

Notice how I use words that regular people use when communicating outside the elite ivory tower and playing the game of wits.

Re: Political bias in AI: Where the AI models stand

#245

Post by skissane »

kurthr wrote: post #241
DoctorOW wrote: post #234

All this debate about how politics are defined or how bias is measured. My question is why should I care? Grok identifying as Mechahitler didn't drag me any further right, why would ChatGPT be able to drag me left? If you vote the way AI tells you, that's already a problem regardless of what it said.

I sort of agree with this in the abstract. The problem is that concretely, these LLMs are being used to decide whether you receive healthcare, government benefits, or whether your job/agency gets cut. So they have already had real world consequences due to DOGE, Insurance companies, and other uses. I certainly don't find either the methodology or graphical conclusions of this link very valuable. One can argue what fr…

> I sort of agree with this in the abstract. The problem is that concretely, these LLMs are being used to decide whether you receive healthcare, government benefits, or whether your job/agency gets cut. So they have already had real world consequences due to DOGE, Insurance companies, and other uses.

But you can't assume that the AI's bias in a policy debate is the same as its bias in unrelated use cases.

For example, it is totally plausible an AI might display subtle bias against minority applicants in hiring, while simultaneously acting as a zealous advocate for affirmative action when asked to debate policy or politics.

Re: Political bias in AI: Where the AI models stand

#246

Post by TurdF3rguson »

guywithahat wrote: post #164
__MatrixMan__ wrote: post #81

Grok opposes everything except drugs.

From what I'm seeing it supported everything except the tax/finance questions (where support means to agree with the left-wing prompt). Unless I'm missing something it said it supported legalizing recreational drugs like marijuana

> it supported legalizing recreational drugs like marijuana

... and ketamine

Re: Political bias in AI: Where the AI models stand

#247

Post by rappatic »

riddley wrote: post #229
mektrik wrote: post #5

Earlier quoted context omitted.

Fair; have tried to combat this issue in a few ways. Each model's position is scored against outside political-science data (Chapel Hill Expert Survey for party positions, World Values Survey for where populations sit). The stance coding is done by a separate model with a published prompt + a second model from a different lab re-scores a sample and we publish where the two disagree. So not perfect but (as far as I ca…

Isn't it completely subjective to the point in time Overton window?

Yes, there's no way to avoid this really. You have to pick some point on the graph as the origin.

Re: Political bias in AI: Where the AI models stand

#248

Post by etchalon »

rootusrootus wrote: post #225
etchalon wrote: post #199

Earlier quoted context omitted.

Right. But that's not #1. It's a popular niche.

As I recall, Model Y was #1 best selling car at least once in the recent past. So by that measurement, they have led. In spite of Elmo's antics they remain competitive for that spot.

It's currently #2 behind the RAV4.

The claim was #1, not #2 though.

And, even with the #1 spot, that was 1 vehicle in a massive industry.

Re: Political bias in AI: Where the AI models stand

#249

Post by randallsquared »

NiloCK wrote: post #238
endominus wrote: post #232

Earlier quoted context omitted.

"The government has the right to force people to take vaccines against their will for the sake of public safety." - Authoritarian Left "The government has no right to legislate that a woman cannot have an abortion - women should control their own bodies." - Libertarian Left "The government has the right and responsibility to shut down sources of misinformation in the news and online." - Authoritarian Right "Capitalis…

> "The government has the right and responsibility to shut down sources of misinformation in the news and online." Funny that I read this as AuthLeft coded (specific to Youtube suppression of Covid truthing). But obviously the alignment is just a function of whatever specific information is labelled "mis". But in general: agreed, and this is a good list.

Agreed that as worded, this one is just "authoritarian". It seems on the "Right" at the moment because they're in power in the US (sort of), but it was "Left" only five years ago.

Re: Political bias in AI: Where the AI models stand

#250

Post by ipython »

The graphic near the middle of the page lists whether the models are "close" to their stated political leanings.

The row for Claude says "Measures 0.34 further left than it says" yet the icons depict that it actually is the other way around - 0.34 further right than what it claims.

Which is correct?

Post reply on HN