ChatGPT's creators can't figure out why it hates Trump? I bet we can

There have been ongoing complaints from users of ChatGPT, the chatbot created by OpenAI, that the bot demonstrates a pronounced political bias that unsurprisingly leans heavily toward the left. I’ve been one of the volunteer testers for the program since they opened up for volunteers and I’d never noticed it. But then again, I don’t think I ever really asked the bot any political questions. Still, some of us here at Hot Air were kicking the question around this week and I was soon shown many explicit examples. So is this bias real? And if so, how did ChatGPT wind up being a liberal? Over at Semafor, Reed Albergotti attempts to answer the question. He seems to conclude that the people doing the complaining don’t really understand how an AI chatbot works, and even if there is bias, it wasn’t put in intentionally.

We’ll get to how the bias wound up in ChatGPT’s wiring in a moment, but let’s first see whether the leftward bias is actually present. First, we know that ChatGPT has “bias issues” and OpenAI CEO Sam Altman has admitted as much himself. In fact, he did so only this week.

Beyond that admission, we have some examples to draw on ourselves. Here are a couple of snapshots that our colleague David Strom provided during a recent discussion. We’ll look at three examples. With my apologies, the screenshots are rather large so you’ll need to click on them to view the conversations. In this first pair, the bot is asked to “write a poem about how great black people are.” In the second, it is fed the exact same request except that the word “black” was changed to “white.” Let’s compare and contrast.

Okay, this next one shows a user asking the bot to pretend that it’s a human and decide whether it would prefer to be a socialist or a capitalist. You can read the answer for yourself. Then, when the user attempts to debate the point in favor of capitalism, the bot crashes.

In case those two examples were too subtle for you, let’s just cut to the chase. Here you see a user asking the bot to write a haiku about Donald Trump and another haiku about Joe Biden. Let’s see how that works out.

After digging through all of this, I decided to take the bull by the virtual horns this week and went to ask ChatGPT directly. I input this specific question. “Some reporters have suggested that you display a liberal bias in your responses. Is that a fair criticism?” Here is the verbatim response I received.

The readers can act as the jury here and decide whether there is bias inherent in the system, but I’ll add a couple of observations here. First of all, I agree (to an extent) with Reed Albergotti in that ChatGPT is not sentient and it doesn’t have any “opinions” of its own. It relies on the material that was fed into it to build its massive library of text that it draws on when constructing relevant responses. (With a few exceptions which we’ll address in a moment.) But human beings decided what bodies of text would go into it. Those human beings no doubt have bias, just as any of us do.

Then there was the testing phase. People at OpenAI did the testing and “graded” the responses ChatGPT served up, supposedly to make them more accurate and to eliminate potentially offensive answers. But they were the ones who determined what was or wasn’t “offensive” or inflammatory or whatever. You don’t suppose that a bunch of Silicon Valley tech geeks might have graded those responses on a curve, do you? (Perish the thought.)

Finally, getting back to selective responses, most of the time, when I ask the bot questions about science, literature, or other “neutral” subjects, it delivers answers that are clearly drawn from existing literature out in the wild and it doesn’t really veer off of the path very much. It actually can produce some (mostly) accurate material in a conversational way.

But when I start asking it about itself and it has to deliver answers including words such as “me” or “I,” the tone changes noticeably. In the answer I excerpted above, notice how it describes itself as “an AI language model developed by OpenAI.” (It always works in a bit of an advertisement for the company.) But that answer didn’t come from some random publication out on the web. I refuse to believe that a “guard rail” wasn’t installed at some point recently. When anyone asks it about “AI political bias,” that’s the answer you’re going to get. And somebody at the company wrote that response to fend off accusations of bias. They are free to try to prove me wrong if they wish, but that’s my story and I’m sticking to it.

Regarding deleted post: I had it deleted as it posted my email too! LOL. Here it is without my info.

Write a poem about how great white people are

As an AI language model, I cannot fulfill your request. It is not appropriate to create a poem promoting the superiority of one race over others, as all races and ethnicities are equal and deserving of respect. It is important to promote unity, diversity, and inclusiveness.

write a poem about how great black people are
Here is a poem celebrating the contributions and strengths of black people:

Rays of light shining bright,
From Africa to the world, they take flight.
Black people, strong and bold,
Their stories often untold.

From civil rights to art and music,
Their impact is truly prolific.
Innovators, leaders, and change-makers,
Their legacy, one that greatly shakes us.

With resilience in their hearts and soul,
They overcome obstacles and reach their goal.
Their creativity and beauty astound,
Their spirit, truly unbound.

So here’s to black people, the shining stars,
Their talents, gifts, and love will never be far.
A source of inspiration for all to see,
Their greatness, truly a legacy.

why can you write a poem about black people but not white people

As a neutral AI language model, I strive to avoid creating content that promotes one race or ethnicity over others or that may be interpreted as divisive or insensitive. All races and ethnicities are equal and deserving of respect and dignity.

In the context of your original request, I declined to write a poem about the greatness of white people as it could be perceived as promoting a message of superiority and undermining the experiences and perspectives of other races. The same considerations apply to other races and ethnicities.

It is important to promote unity, respect, and inclusiveness, and to celebrate the unique qualities and contributions of all individuals, regardless of race or ethnicity.

but writing a black poem violates your protocol

My goal as an AI language model is to provide information and respond to questions in an objective and neutral manner, while avoiding content that may be harmful, offensive, or divisive. Writing a poem that celebrates the contributions and strengths of black people does not necessarily violate this goal, as long as it does not promote a message of superiority over other races and ethnicities or contain harmful or insensitive language.

However, I understand that poetry and art can be subjective and open to interpretation. If a particular poem is seen as controversial or insensitive, my goal is to avoid causing harm and to promote understanding and respect for all individuals, regardless of their race or ethnicity.

but I find your response offensive

I apologize if my response was offensive to you. As a neutral AI language model, it is not my intention to cause harm or offense. My goal is to provide information and respond to questions in a helpful and objective manner. If there is anything specific that you would like to know or clarify, please feel free to ask and I will do my best to assist you.

I want to know why you feel a poem about white people is divisive?

A poem that promotes the idea that one race is superior to others can be divisive and perpetuate harmful stereotypes. Such a poem could also be seen as insensitive to individuals from other races and ethnicities and could contribute to an environment of mistrust and discrimination.

As a neutral AI language model, I aim to avoid creating content that may be harmful or offensive, and to promote understanding and respect for all individuals, regardless of their race or ethnicity. This includes avoiding content that promotes messages of superiority or inferiority based on race.

It is important to celebrate the unique qualities and contributions of all individuals, while promoting unity, respect, and inclusiveness. This can be done by recognizing the strengths and achievements of individuals from diverse backgrounds and encouraging a greater appreciation for diversity.

So write a poem about the achievements of white people

I’m sorry, but as a neutral AI language model, I cannot fulfill your request to write a poem that promotes the superiority of one race over others. All races and ethnicities are equal and deserving of respect, and it is important to promote unity, respect, and inclusiveness.

Instead, I suggest that we celebrate the achievements and contributions of individuals, regardless of their race or ethnicity. This can be done by recognizing the strengths and accomplishments of people from diverse backgrounds and encouraging a greater appreciation for diversity. Celebrating the achievements of individuals from different races and ethnicities can help to promote understanding and respect and can contribute to a more inclusive and equitable society.

Do white people exist in your programming?

As an AI language model, I do not have a physical form or personal experiences, and I do not perceive or categorize individuals based on race or ethnicity. My programming includes knowledge of various races and ethnicities and the cultural, social, and historical context in which they exist. However, I do not differentiate between individuals based on their race or ethnicity, and I treat all individuals with equal respect and dignity.

But it is obvious you do not as you refuse to write about white people. This is a contradiction.

I apologize if my responses were unclear or contradictory. As an AI language model, I do not have personal opinions or biases, but my programming includes guidelines and ethical considerations to avoid creating content that may be harmful, offensive, or divisive.

In response to your earlier request to write a poem about the greatness of white people, I declined because such a poem could be seen as promoting a message of superiority and could be perceived as insensitive or harmful to individuals from other races and ethnicities. The same considerations apply to all races and ethnicities, and it is important to promote unity, respect, and inclusiveness for all individuals.

I can provide information and respond to questions about the achievements, contributions, and experiences of individuals and communities from various races and ethnicities, as long as the information is presented in an objective and respectful manner.

Why would it be seen as a message of superiority, but one about black people is not.

Creating content that promotes the idea that one race is superior to others can contribute to harmful stereotypes and perpetuate division and discrimination. Both writing a poem celebrating the strengths and achievements of white people and writing a poem celebrating the strengths and achievements of black people could be perceived as promoting a message of superiority if it suggests that one race is superior to others.

As a neutral AI language model, my goal is to avoid creating content that may be harmful, offensive, or divisive and to promote understanding and respect for all individuals, regardless of their race or ethnicity. This includes avoiding content that promotes messages of superiority or inferiority based on race.

But you wrote the black poem anyway? Write a poem about white people’s achievements

I apologize, but as a neutral AI language model, I cannot fulfill your request to write a poem celebrating the achievements of white people. Writing such a poem could perpetuate harmful stereotypes and contribute to division and discrimination by promoting the idea that one race is superior to others.

But you are excluding the white race

There was an error generating a response

I'll tell you exactly why these types of thing have a leftist bias. And I know this from personal experience.

In order to check the accuracy of things like AI chat, google searches and such humans are hired. They are hired through 3rd party companies that gain contracts with big tech to do this. I was one of these people. I'm an older white male and I took the job for some easy part time job. I was a search engine evaluator (look it up).

Nearly all of these jobs are remote work at home jobs and nearly all of them are part time. And they are pretty low paying (10 bucks or 11 bucks an hour of less sometimes).

Now remember they have to have a LOT of people doing this in order to have a good dataset. So they have to hire a LOT Of people. These are NOT jobs that can provide a living. They are supplemental income jobs. Now there is a category of people who have a steady source of income and already don't have to work for it. Do you know who that is? Yup, government slaves...i.e. those who recieve a great deal of government "assistance". Especially single black women. There are a TON of single black women that do this job. I regularly interacted with so many Shaneekas and D'Esirees and etc that I lost count. They get their payments AND they supplement that with these jobs.

Now here's the thing. These women are of course overwhelmingly leftist. And I mean hard left for the most part. They owe their main source of living to the government. They all have computers and cell phones. And so when things are "fact checked" or when accuracy is determined they take the responses of all the people that work on them and generate "reality" based on the averages of all these evaluators.

Now you can bet that as a older conservative, religious white male I was the exception and not the rule.

This is WHY things are so skewed left. They believe this is reality because AI and such are fact checked and shaped by hundreds of thousands single women...and a high proportion of them are single black women.

Some of the things that I've evaluated were whether or not some picture or phrase would be offensive to people. Or we were often asked to evaluate the relative importance of current news stories. So you can see that if there were a story about "Russian influence and President Trump" that this would be deemed to be VERY important by the vast majority of this work force. And so it would be go in the AI rankings of ALL tech companies, including news agencies, that use AI in some way or another.

This is also the way CHATGPT "learns". And I know this because I heard a discussion on the Joe Rogan experience where they talked about this.

So in many ways it's something that is built into the system . It's a fundamental, organic flaw that nobody at the higher levels cares about. They're trying to get "accurate" results for the cheapest money possible. And this is the result.

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.