Kerby Anderson

Is there a leftist bias in artificial intelligence? A Washington Post tech writer has written about various research papers about the liberal bias within Open AI’s ChatGPT. Although the AI program tells users that it doesn’t have any political opinions or beliefs, it shows certain biases.

A study by the Springer Journal of Public Choice also found that ChatGPT produced responses that were in line with leftist thought. The Massachusetts Institute of Technology found that OpenAU’s models typically produce left-leaning responses.

One extensive study came from John Lott, who has been on my radio program many times in the past. His Crime Prevention Research Center asked questions of twenty AI chatbots on topics related to crime and gun control. He then ranked the answers on how liberal or conservative their responses were.

Only Elon Musk’s Grok AI chatbots gave conservative answers on crime. But even these programs gave consistently liberal answers on gun control. Bing was the least liberal chatbot on gun control. The French AI chatbot Mistral in the only one that on average was neutral in its answers.

John Lott concludes that these AI programs are very liberal in their answers on crime and gun control. But they also provide liberal answers on economic and social issues, with Google’s Gemini being the most extreme. He also notes that Musk’s Grok has moved more toward the political center, but much more needs to be done.

As I document in my recent booklet on A Biblical View on Artificial Intelligence, there are many concerns that researchers have about AI, but one that certainly needs to be addressed soon in the obvious leftwing bias in these programs.