I recently found myself arguing with artificial intelligence.
Now, there’s a sentence I never thought I’d write.
And before you start wondering whether I’ve completely lost my mind, let me explain. I wasn’t arguing with my computer because it wouldn’t turn on, or because I couldn’t remember a password. I was arguing with ChatGPT over something that ought to have been fairly straightforward.
I wanted some statistics.
Specifically, I wanted to know what percentage of Baby Boomers support the Second Amendment and how their opinions compare with those of younger generations.
Simple enough, right?
Well, not exactly.
When a Simple Question Gets Complicated
ChatGPT came back with an answer, complete with percentages, explanations, and references to polling organizations.
It looked pretty convincing.
But I wasn’t particularly satisfied with the sources it selected. So I challenged them and asked for information from other polling organizations.
And wouldn’t you know it? Additional research turned up more information, including some interesting differences in how the questions were asked and how the results were reported.
Then I noticed something else.
Some of the information was several years old. Here we were in 2026, and my supposedly sophisticated research assistant was relying heavily on polling from 2022.
So I challenged that, too.
More recent surveys were available, and ChatGPT eventually produced an updated report.
Problem solved?
Not quite.
I continued questioning its choice of sources, including some with media affiliations I personally don’t consider particularly balanced.
By this point, I was becoming more than a little frustrated.
I had repeatedly instructed ChatGPT to give me balanced information, examine different sources, and avoid approaching controversial subjects from a predetermined political perspective.
Yet I kept finding myself pushing it to broaden the research.
And that brings me to something I’ve been wondering about for quite some time.
Is Artificial Intelligence Really Unbiased?
I have increasingly come to suspect that artificial intelligence, or at least the version I’ve been using, has a built-in liberal bias.
I can’t tell you whether that comes from its original programming, its training, or the information sources it favors.
But I can tell you what I’ve experienced.
When I ask about politically controversial subjects, ChatGPT repeatedly seems to start with sources and perspectives associated with the political left. Only after I challenge those choices does it broaden its research to include more conservative perspectives and alternative sources.
And here’s what really bothers me.
I’ve specifically instructed it to provide balanced, unbiased information, yet the pattern keeps repeating.
I’m not asking it to give me conservative answers. I’m asking it not to automatically favor liberal perspectives, either.
I want the facts. I want credible sources. I want competing viewpoints examined fairly. And I want enough information to reach my own conclusions.
Is the recurring pattern deliberate programming? A consequence of how artificial intelligence is trained? Or something else entirely?
I don’t know.
And to be fair, I can’t prove that ChatGPT has been deliberately programmed to favor a particular political ideology.
But I don’t have to prove that to question what I’m experiencing.
I think it’s a legitimate concern, and I don’t intend to stop asking questions about it.
Because whether the bias leans liberal or conservative, the problem is the same.
Information presented through a political filter isn’t the same as an honest examination of the facts.
And I don’t believe anyone should have to fight their research assistant to get the whole picture.
Here’s What Really Bothered Me
It wasn’t simply that I disagreed with some of the sources.
It was that I had to keep challenging the research to get a more complete picture.
Think about that for a moment.
What happens when someone asks an AI system a question, gets a polished answer with several impressive-looking citations, and simply accepts it?
Most of us aren’t going to investigate every statistic or look up every research organization.
We’ve got lives to live.
And that’s precisely why this matters.
Artificial intelligence has an extraordinary ability to present information in a manner that sounds confident, organized, and authoritative.
But sounding authoritative and being correct are two different things.
A well-written answer isn’t necessarily a well-researched answer.
And a citation isn’t some magical guarantee that the information is accurate, complete, or presented without bias.
Sometimes the Question Matters as Much as the Answer
Something interesting came out of my little disagreement with ChatGPT.
We discovered that polling about gun rights can produce substantially different results depending on what people are actually asked.
For example, asking whether someone supports repealing the Second Amendment isn’t the same as asking whether they support additional restrictions on firearms.
Someone can strongly support the constitutional right to keep and bear arms while also supporting certain gun regulations.
Those positions aren’t necessarily contradictory.
That’s important information.
And it’s exactly the kind of distinction that can get lost when we’re handed a percentage without understanding the question behind it.
The same thing happens with politics, economics, public health, national security, and just about anything else people argue about these days.
Sometimes we’re disagreeing about the facts.
Other times we’re looking at different pieces of the same picture.
And sometimes the information being presented really is incomplete, outdated, or misleading.
The trick is knowing the difference.
Don’t Just Challenge What You Disagree With
Now, here’s where I think we all need to be careful, myself included.
It’s easy to question information when it contradicts something we believe.
But what about information that supports our existing opinions?
Do we scrutinize that just as carefully?
Probably not as often as we should.
If an AI system tells me something that lines up perfectly with my views, I ought to be just as interested in whether the information is accurate as I would be if it told me something I didn’t want to hear.
Otherwise, I’m not really researching anything.
I’m just looking for somebody, or something, to agree with me.
And that’s not particularly useful.
A Few Questions Worth Asking
You don’t have to be a professional researcher to challenge information you find online.
Start with a few common-sense questions.
Who produced the information?
How old is it?
What was the actual question being asked?
Are there other credible sources reporting something different?
And perhaps most importantly, is there information missing that might change how I understand the answer?
Those questions apply whether you’re reading a news article, watching a YouTube video, scrolling through social media, or asking ChatGPT to research something.
Don’t assume the first answer is the complete answer.
And don’t be afraid to say, “Show me where you got that.”
A Powerful Tool, But Still a Tool
Now, I don’t want anyone getting the wrong impression.
I’m not opposed to artificial intelligence. Quite the contrary.
I use ChatGPT regularly, and I’ve found it tremendously useful for research, writing, organizing information, and tackling projects that would otherwise take considerably more time.
It’s an impressive piece of technology.
But I don’t believe impressive technology should be confused with unquestionable authority.
I don’t want AI deciding what I ought to believe. I want it helping me find and understand the information I need to make that decision for myself.
There’s a big difference.
And I don’t think I should have to argue with a computer to get an honest, balanced examination of an issue.
Of course, there’s a certain amount of irony in the fact that I asked ChatGPT to help me develop an article about why people shouldn’t automatically trust ChatGPT.
I haven’t quite figured out whether that makes me clever or just another willing participant in the experiment.
Probably a little of both.
But here’s the difference.
I’m reading what it writes, questioning what it tells me, and deciding for myself what belongs in this article.
And if something doesn’t sound right, you can bet I’ll challenge it.
The Bottom Line
We’re living in a time when information is available faster and in greater quantities than at any point in human history.
That doesn’t necessarily mean we’re better informed.
Technology can help us find answers, but it can’t relieve us of the responsibility to think critically about those answers.
So use the tools. Take advantage of what they offer. Ask questions and explore subjects you might never have investigated otherwise.
But don’t surrender your judgment in the process.
Challenge the answers. Check the sources. Look for other perspectives. And don’t mistake confidence for accuracy.
Because whether the information comes from a television screen, a website, a social media account, or an artificial intelligence system, the responsibility for deciding what we believe still belongs to us.
And as for ChatGPT?
I’ll keep using it.
But I’ll also keep questioning it.
After all, somebody has to keep the computer honest.