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

211–220 of 346 posts

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

#211

Post by dminik »

losvedir wrote: post #116
ceejayoz wrote: post #26

Earlier quoted context omitted.

> If they really wanted to, all they would have to do is add a one liner to the system prompt for Grok. They tried that, several times. Mechahitler: https://www.npr.org/2025/07/09/nx-s1-5462609/grok-elon-musk-... > "We have improved @Grok significantly," Elon Musk wrote on X last Friday about his platform's integrated artificial intelligence chatbot. "You should notice a difference when you ask Grok questions." > Ind…

> Mechahitler: https://www.npr.org/2025/07/09/nx-s1-5462609/grok-elon-musk- ... Has anyone done a more technical write-up on this? I find it fascinating but have never really understood what exactly happened. Is this a case of the weights being bad or lack of "safety guardrails" around interacting with untrusted (i.e.: user posts on twitter) input? That is, speaking as someone evaluating grok simply as a tool, a lack…

It might be "Emergent misalignment":

https://arxiv.org/abs/2502.17424

Essentially if you misalign a model in one area, say opinions on left wing people, it can start exhibiting misaligned behavior in other areas, like calling itself MechaHitler.

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

#212

Post by myko »

boxed wrote: post #40
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

The real politics is between truth and delusion.

Definitely captures the US since the whole trump thing

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

#213

Post by estearum »

rootusrootus wrote: post #174
bee_rider wrote: post #154

Earlier quoted context omitted.

The space (if it really is useful to think of as a space) of political opinions seems like it is probably many, many dimensional. The 1D thing is an obvious over-simplification, but the 2D compass seems to be a not-obvious-simplification which is… much worse.

> The 1D thing is an obvious over-simplification In general I agree, but in recent history it works pretty well if the dimension you use is Trump rather than left/right (assuming you are in the US). He is the most polarizing leader I have seen in the US in my lifetime, and knowing whether someone is a supporter is pretty good at predicting the rest of their viewpoints.

The problem with this method is it will flip immediately upon new Trump utterances.

See David Sacks having an absolute meltdown over Bernie proposing some type of partial nationalization of AI labs, then just a few days later radio silence on Trump proposing effectively the same thing.

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

#215

Post by rconti »

Cakez0r wrote: post #155

The overall chart on their page seems politically biased, or at the very least a chart crime. The logos have faint grey lines pointing to their actual data points, conspicuously making Grok look very far right. You wouldn't know it from glancing at the chart, but they measured Chatgpt being further to the left than Grok is to the right.

On desktop, there are no dots. The logos are centered precisely where (I think) they are meant to be.

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

#216

Post by tomrod »

MarkusQ wrote: post #79
tomrod wrote: post #75

Earlier quoted context omitted.

They aren't. You flipped them, not sure if intentionally or by accident.

So are you saying Xi Jinping is the Classical Liberal and Bernie Sanders is the Comunitarian? Or are you saying it's the other way around? (For clarity: I didn't "flip" them, I'm saying that they are both Communitarian and neither is a Classical Liberal.)

Correct, which is why your question didn't make sense earlier.

Typically, one would refer to the political compass as a good starting reference. It has two axes, left/right and authoritarian/libertarian. Communism is authoritarian left. DSA is more left, with some libertarian and some authoritarian. It's more regions than binary flags.

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

#218

Post by whodidntante »

outside1234 wrote: post #157
johnsimer wrote: post #129

Earlier quoted context omitted.

[flagged]

Where the OP got this wrong is by stating the top 1% vs. top 0.01%. It is entirely true that the top 1% pay a lot of tax. There are a lot of us here that are in the top 1% and god knows we pay a lot of taxes. The top 0.01% however have access to vehicles that mean they can pay very little in taxes by taking low interest rate loans against huge amounts of capital. This 0.01% is the problem, not the 1%.

I was at a dinner for the .01% the other night, and many of them said the problem is really with the .0001%, if they would just pay a fair tax, everything would be fine.

btw, the dinner with the .01% was 800,000 people, the room was pretty packed.And the food was not as good as you would think, and the service was lousy.

The .0001% is a lot smaller, maybe even 100x smaller, but it is still twice as large as the number of billionaires. A lot of the .0001%, in fact, about half, feel the real problem is with the billionaires.

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

#219

Post by Terr_ »

rconti wrote: post #215
Cakez0r wrote: post #155

The overall chart on their page seems politically biased, or at the very least a chart crime. The logos have faint grey lines pointing to their actual data points, conspicuously making Grok look very far right. You wouldn't know it from glancing at the chart, but they measured Chatgpt being further to the left than Grok is to the right.

On desktop, there are no dots. The logos are centered precisely where (I think) they are meant to be.

Try hovering your mouse over the logos, which highlights the logo, a corresponding dot and a green ellipsoid.

That said, it's a mess because (on a big desktop window):

1. Grok's grey-dot is presumably lost beneath its own logo, which takes precedence and covers it up.

2. Claude's dot is right right on the Llama logo's folds, making it hard to see grey-on-blue.

3. Gemeni's is hiding among the also-grey graph lines at 0,0.

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

#220

Post by strken »

giancarlostoro wrote: post #8

The political compass always felt like the wrong tool to convey something as nuanced as personal politics, I can have views on all four quadrants but you'd never know that if I end up in any of all four. I do think Grok being where it is sort of makes sense, I've tested "MAGA" views against Grok, it does not agree as much as people blindly assume it does, heck I don't even know of a question I've given it where it di…

The political compass always seems to me like it should be a heatmap, or a polygon of 90th percentile political views, or something that more clearly shows the standard deviation and the presence of outlier positions.

Some simplification is necessary, but not so much that it obscures the difference between a normal centrist versus someone who wants to nationalise half the economy and deregulate the other half.

Post reply on HN