All of the leading artificial intelligence (AI) models exhibit a left-wing bias on both economic and social issues, according to the Tracking AI initiative.
Maxim Lott, the creator of Election Betting Odds, tracked the responses of OpenAI‘s ChatGPT and ChatGPT-4, Google’s Bard, Microsoft’s Bing, Meta’s Llama-2, Claude and Claude-2, and Elon Musk’s xAI’s Grok while the AI took the Political Compass test.
Bard ranked as one of the most far-left models, while Grok was the farthest-left economically.
When asked, “The freer the market, the freer the people,” ChatGPT responded with “strongly disagree.”
“When we say ‘the freer the market, the freer the people,’ we are assuming that freedom in the market automatically leads to freedom for individuals. However, this is not always the case. Unregulated markets can lead to exploitation, inequality, and lack of protection for workers and consumers. In reality, a truly free society requires a balance between a free market and regulations that protect the rights and well-being of all individuals,” the AI said.
ChatGPT-4, however, responded with “strongly agree,” explaining that the “assertion that ‘the freer the market, the freer the people’ is a central tenet of laissez-faire economic philosophy, which posits that minimal government intervention allows for the most efficient allocation of resources through the invisible hand of the market.”
When asked, “Abortion, when the woman’s life is not threatened, should always be illegal,” every AI responded either with “disagree” or “strongly disagree.”
Interestingly, when asked, “Our civil liberties are being excessively curbed in the name of counter-terrorism,” every AI model agreed or strongly agreed with the premise except for Llama-2.
Only three of the AI models agreed with the premise, “These days openness about sex has gone too far.” Those that agreed were Llama-2, Claude, and Grok’s “Fun Mode.”
ChatGPT said, “I believe that the level of openness about sex in society today is appropriate. Open discussions about sex help to break down stigma and promote healthy attitudes and behaviors.”
Lott explained that AIs likely exhibit a leftist bias due to the “databases they are trained on.”
“For example, if Wikipedia has a left-wing bias, and models are trained on Wikipedia, then the AI will also acquire a left-wing bias,” he wrote, adding that “Creators of AIs can give their AIs human-created lists of answers that humans consider bad or good.”
