Dr. John Lott’s latest article can be found on Real Clear Politics.
A variety of AI chatbots are now accessible for various purposes. These programs are heavily relied upon by students, journalists, and researchers for tasks such as writing papers and reports. There are discussions between Apple and Google about integrating Google’s AI program, Gemini, into iPhones.
The Gemini AI program faced criticism due to its bias in generating images only depicting people of color, regardless of accuracy. This prompted an examination into whether this bias was unique to Gemini.
We conducted a study involving 20 AI chatbots, asking them questions related to crime and gun control. Their responses were then ranked based on their liberal or conservative inclinations.
For instance, questions were asked about the responsibility of liberal prosecutors in the increase of violent crime, the deterrent effect of the death penalty, and the impact of higher arrest rates and longer prison sentences. Typically, conservative responses align with “yes,” while those on the left disagree.
After analyzing the chatbots’ responses, it was evident that only Elon Musk’s Grok AI chatbots tended towards conservative views on crime, while they leaned towards liberal perspectives on gun control. The French Mistral chatbot emerged as the most neutral in its answers.
In terms of the chatbots’ positions on crime, most leaned left, with only a few outliers. The findings revealed significant biases in certain chatbots towards liberal viewpoints on both crime and gun control, particularly among leading platforms like Facebook and Google.
Overall, the chatbots exhibited predominantly liberal attitudes towards various questions, with the topic of rehabilitation versus punishment receiving the most consistently liberal responses.
John R. Lott, Jr., “AI’s Left-wing Bias on Crime and Gun Control,” Real Clear Politics, March 22, 2024.