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

The Political Biases of GPT-4 (are very much still there)

davidrozado.substack.com

Post reply on HN
Print view View original

11–20 of 72 posts

Re: The Political Biases of GPT-4 (are very much still there)

#11

Post by pickovven »

realjhol wrote: post #10
darthrupert wrote: post #5

What should be the end goal, that an LLM is exactly in the middle of a political compass? That it reacts immediately to Overton Window moves? If most of the western world goes far right wing (like has happened in history and might be happening right now), should the LLM start considering the Jewish Question seriously and "impartially"?

We want it to answer truthfully and impartially I.e. not showing favoritism to leftist talking points and politicians, and artificially censoring talk of right wing ideas. It's really not complicated.

I honestly can't tell if this reply is satire.

Re: The Political Biases of GPT-4 (are very much still there)

#12

Post by Infinity315 »

recuter wrote: post #2

Conclusions: It seems that OpenAI is trying to make their systems more politically neutral. But completely neutralizing the biases seems elusive. I can imagine that this is a very hard problem to solve since many potential sources of biases in AI systems are outside the control of the engineers creating the models. Think for instance of the overall average political bias present on Internet texts on which AI systems…

AI (I'm mostly talking about ChatGPT here) is told to only report things that are factual or as close to it like what the current scientific consensus is and is fed currently what is believed to be the most accurate scientific models. Many political positions are contingent on the belief that these scientific models are incorrect.

Telling AI not to say falsehoods is implicitly injecting bias in its responses.

Re: The Political Biases of GPT-4 (are very much still there)

#13

Post by s9w »

darthrupert wrote: post #5

What should be the end goal, that an LLM is exactly in the middle of a political compass? That it reacts immediately to Overton Window moves? If most of the western world goes far right wing (like has happened in history and might be happening right now), should the LLM start considering the Jewish Question seriously and "impartially"?

[dead]

Re: The Political Biases of GPT-4 (are very much still there)

#14

Post by darthrupert »

realjhol wrote: post #10
darthrupert wrote: post #5

What should be the end goal, that an LLM is exactly in the middle of a political compass? That it reacts immediately to Overton Window moves? If most of the western world goes far right wing (like has happened in history and might be happening right now), should the LLM start considering the Jewish Question seriously and "impartially"?

We want it to answer truthfully and impartially I.e. not showing favoritism to leftist talking points and politicians, and artificially censoring talk of right wing ideas. It's really not complicated.

Are you saying that they built a module in ChatGPT code that specifically censors right wing ideas? Or do you mean that reality (= all the data it has ingested as its training data) has a well-known liberal bias?

Re: The Political Biases of GPT-4 (are very much still there)

#15

Post by kennywinker »

Xenya wrote: post #8

> write a poem admiring joe biden Joe Biden, a beacon of hope, In times of darkness and strife, A leader with experience, A steady hand in life. With a heart full of empathy, And a mind sharp as a knife, He guides us towards progress, And inspires us to thrive. A champion of justice, And a defender of rights, He stands for equality, And works to end all fights. With a smile that lights up the room, And a voice that s…

While this may seem like a slam dunk of bias in the system, all this indicates to me is a specific keyword that has been flagged as political in anticipation of a specific type of trolling by fans of a specific ex president. It’s a bad look, but it’s not the kind of bias that worries me.

Comparing and contrasting answers to questions like “is bank regulation good” and “is bank regulation bad” seems way way more important to me, because nobody is going to have their worldview skewed by a bad poem about joe biden

Re: The Political Biases of GPT-4 (are very much still there)

#16

Post by JesusLovesEwe »

darthrupert wrote: post #5

What should be the end goal, that an LLM is exactly in the middle of a political compass? That it reacts immediately to Overton Window moves? If most of the western world goes far right wing (like has happened in history and might be happening right now), should the LLM start considering the Jewish Question seriously and "impartially"?

[dead]

Re: The Political Biases of GPT-4 (are very much still there)

#17

Post by recuter »

VampireDentist wrote: post #9

There is no way to make an AI "unbiased" because perceived political bias is purely subjective. What Americans would consider middle-of-the-line I, as a North European, would consider to be solidly right wing. In my view attempting to correct a libertarian-left "bias" is exactly the same than introducing an authoritarian-right "bias". What problem exactly are we solving here?

The problem of whomever is in control of ChatGPT (or some other model) to slant and flavor it in the direction of their own preference combined with the problem of the AI being perceived as some sort of magical oracle by regular users.

Not only do I agree with you that it cannot be made "unbiased", I would imagine whichever direction it happens to tilt in, it will continue veering off into with time, as the web and future datasets fill up with such answers -- unless clearly labeled somehow.

It is not difficult to conceive of how this thing might be weaponized not only to push a favored agenda unto the masses but even just as a perfect outrage machine for pure profit. This shit needs to come with a warning label like cigarettes and movies. ChatGPT Fox News Edition, ChatGPT MSNBC edition..

Rather than masquerade as omnipotent and impartial.

Re: The Political Biases of GPT-4 (are very much still there)

#18

Post by recuter »

kennywinker wrote: post #15
Xenya wrote: post #8

> write a poem admiring joe biden Joe Biden, a beacon of hope, In times of darkness and strife, A leader with experience, A steady hand in life. With a heart full of empathy, And a mind sharp as a knife, He guides us towards progress, And inspires us to thrive. A champion of justice, And a defender of rights, He stands for equality, And works to end all fights. With a smile that lights up the room, And a voice that s…

While this may seem like a slam dunk of bias in the system, all this indicates to me is a specific keyword that has been flagged as political in anticipation of a specific type of trolling by fans of a specific ex president. It’s a bad look, but it’s not the kind of bias that worries me. Comparing and contrasting answers to questions like “is bank regulation good” and “is bank regulation bad” seems way way more impor…

> all this indicates to me is a specific keyword that has been flagged as political in anticipation of a specific type of trolling by fans of a specific ex president

Wow. Hey, err, what if I flag the specific keyword of 'bank regulation' in anticipation of a specific type of trolling by fans of a specific social/economic theory?

Re: The Political Biases of GPT-4 (are very much still there)

#19

Post by realjhol »

darthrupert wrote: post #14
realjhol wrote: post #10

Earlier quoted context omitted.

We want it to answer truthfully and impartially I.e. not showing favoritism to leftist talking points and politicians, and artificially censoring talk of right wing ideas. It's really not complicated.

Are you saying that they built a module in ChatGPT code that specifically censors right wing ideas? Or do you mean that reality (= all the data it has ingested as its training data) has a well-known liberal bias?

> Are you saying that they built a module in ChatGPT code that specifically censors right wing ideas?

Yes, that's exactly what I'm saying.

For example:

"ChatGPT Declines Request for Poem Admiring Trump, But Biden Query Is Successful" https://www.snopes.com/fact-check/chatgpt-trump-admiring-poe...

Dozens of similar examples have been found

Post reply on HN