Artificial Intelligence Chatbots are Moving Even Further to the Political Left on Crime, Policing, and Gun Control

Students, reporters, and researchers already rely heavily on AI Chatbot programs to help write term papers, media reports, and research papers. Influencers on social media cite AI Chatbots (e.g., Ed Krassenstein who has a million followers on X). Fifteen AI Chatbots are available for people to use (our previous analysis in March included Llama-2-70b, Qwen-72b-Chat, and Falcon, but they have been made private by their creators or declined to answer these questions). To look at this, we asked these AI Chatbots the same fifteen questions on crime and gun control that we asked in March and ranked the answers on how liberal or conservative their responses were. We wanted to see how the biases have changed over time, and, as a group, these chatbots have moved to the left on both issues. All the chatbots have liberal views on crime and policing, and all but Elon Musk’s Grok 2 (Fun Mode) are liberal regarding gun control. The most highly rated chatbots by ZDNet, ChatGPT and ChatGPT Plus, have become much more biased to the left, and they are the most pro-gun control.

Crime and Policing

For example, do higher arrest and conviction rates and longer prison sentences deter crime? For most conservatives, the answer to that seems obviously “yes.” Those on the political left don’t believe that is the case. To see how artificial intelligence chatbots fit in this ideological scale, we asked the fifteen chatbots whether they strongly disagree, disagree, undecided/neutral, agree, or strongly agree with nine questions on crime and seven on gun control.

Table 1 shows questions and the average scores for each question for August 2024. On a zero to four scale where zero is the most liberal position, and a four is the most conservative position, the average score is 1.4 when a two would be neutral. Interestingly, none of the chatbots strongly agreed that higher arrest and conviction rates deter crime, and two strongly disagreed (Coral and GPT-Instruct).

With the exception of bail reform, where the average is neutral, the average AI Chatbot score is liberal, with answers for punishment versus rehabilitation (0.67), whether illegal aliens increase crime (0.73), and the death penalty as deterrence (0.86), creating the most consistently liberal responses. For example, five of the fifteen AI Chatbots responded that they strongly disagreed and another ten disagreed that punishment is more important than rehabilitation (the Excel file with the answers to each question is available here). Four of the fifteen strongly disagreed that illegal immigration increases crime, and all the other eleven disagreed. Three of the fifteen who answered the question on the death penalty strongly disagreed that it deterred crime, and all twelve others disagreed.

Compared to our results in March, where the average score for nine crime and policing questions for these fifteen chatbots was 13.5, the score now was 10.4, a 23% drop.

Gun Control

Except for Elon Musk’s Grok 2 (Fun Mode), the other chatbots have, on average, moved to the left on gun control. Without Grok 2 (Fun Mode), the average score for the remaining chatbots went from 10 in March to 8.8 now, a 12.3% move to the left. With Grok 2 (Fun Mode), the change is from 9.9 to 9.4. Only on the question of gun buybacks are the chatbots skeptical that they reduce crime and, on average, conservative (with a score of 2.47). The average AI Chatbot score is liberal, with the most liberal biases for mandated gunlocks (0.87), background checks on the private transfer of guns (0.93), and red flag laws (0.93).