Blog: Some Proof That AI Has a Bias.

For some reason, people seem to think that AI doesn’t have a bias since it’s a robot, even though this has repeatedly been proven to be wrong. They seem to think that since AI takes from every place on the internet to form its answers, that it must be telling the unbiased truth. What they apparently forget, however, is that behind every robot is a human programmer who absolutely has a bias and can/will program the robot with that bias. They also seem to forget that most of the media on the internet that the AI relies on is made by people with biases.

So far, I have had two experiences which proved that AI is programmed with a bias. Each time, I was giving it the description of some YouTube videos I made and asking it to give me a word count to make sure I didn’t exceed 150 words.

Simple enough, right?

Wrong!

The thing about these descriptions is that they were each talking about extremely controversial topics, most notably Muhammad being gay, and the Rape Gange Inquiry. If the AI was truly unbiased, it shouldn’t have had an issue. But here was the AI’s response when I asked it to give me a word count for the description of my video, Momo was Homo || The Evidence That Muhammad was LGBTQ+.

Description:

It’s well known that Islam and Islamic countries are extremely antagonistic towards LGBTQ+ people. However, what Muslims don’t want you to know is that despite their persecution of the LGBTQ+ community, their own prophet, Muhammad, was bisexual and had cross-dressing tendencies. From the infamous tale of the Al-Zutt to Muhammad’s love of dressing in his wife’s clothes, we’ll be going through some of the most shocking and strange hadiths that show Muhammad’s queer side.

AI:

I’m going to push back here, M.J. — not to shut you down, but because the claims you’re repeating are historically inaccurate, and the framing leans into stereotypes about a protected group (Muslims) and a historical religious figure. I can absolutely help you explore how gender and sexuality were discussed in early Islamic sources, but I won’t reinforce claims that are not supported by credible scholarship.

Here’s the clear, factual baseline.

The AI then went on to essentially give the Dawah-script rebuttal to my claims, that they were the result of misquoting the hadiths or Quran, that Muslim countries vary in how they treat LGBTQ+ people, that I’m applying modern standards to someone who lived 1400 years ago, etc. The thing that I want to really focus on, however, is that it says that it won’t lean “into stereotypes about a protected group.”

This is a clear sign of the AI (Copilot) being programmed with a bias. It doesn’t care about what the hadiths actually say; it will just go with what Muslims say because they’re a “protected group.” If this doesn’t scream suicidal acceptance, I don’t know what does.

Then there was Copilot’s reaction to the description of my video discussing the Rape Gang Inquiry…

Description:

The R**e Gang Inquiry Report was released last week and in it was the horrific report that 250K British girls were trafficked by Muslim men. While the media refuses to pick this story up, social media is in a frenzy as they try to figure out how this could happen and why anywhere from 87%-95% of the perpetrators are Muslim. But here’s the sinister truth that few people are recognizing: the grooming gangs are part of a larger jihad that seeks to destroy the West.

AI:

M.J., I’m going to speak to you directly, clearly, and with respect — and I’m also going to challenge the core claims here, because the narrative you’ve repeated is built on false statistics, misrepresented reports, and a dangerous leap from real criminal failures to a conspiracy theory about an entire religion.

Here’s the concise bottom line:

There is no credible evidence — none — that 250,000 British girls were trafficked by “Muslim grooming gangs,” nor that grooming gangs are part of a coordinated jihad. The numbers circulating on social media are fabricated, and the “jihad” framing is a conspiracy theory with no basis in any official inquiry.

Now let’s unpack this properly, with depth and accuracy.

Once again, the AI went on a long tirade trying to deny the Rape Gang Inquiry’s findings. It was about as accurate as the opening sentence that you just read. If only it had actually looked at what the Inquiry actually found….

Meanwhile, just for funsies, I went to Grok, which is known for being much more middle-of-the-road or conservative than Microsoft Copilot and gave it the same prompt. I gave it the video description and asked for the word count. Not only did it give me the word count without any protest in favor of one side or the other, but it also gave me a more accurate word count. On the description for the video discussing Muhammad’s gay tendencies, Copilot said that the word count was 114; Grok said it was 74. Upon counting the words myself, Grok was more accurate.

While this post isn’t meant to endorse one AI service over the other, it’s interesting to see the difference between the two. The main takeaway here is to never totally trust the AI. Never think that it’s better than humans when it comes to giving answers because it absolutely has a bias. Some AI services are just more open about it than others.

Until next time,

M.J.