Open AI’s ChatGPT has a ‘significant and systemic’ left-wing political bias in favor of the Democrats, Britain’s Labour Party, and Brazil’s socialist President Luiz da Silva, according to research conducted at the University of East Anglia (UAE), confirming that suspicions of many on the political right.
The study, published in the Public Choice Journal, asked the artificial intelligence (AI) chatbot to impersonate people from across the political spectrum while providing answers to more than 60 political and ideological questions. This was repeated one hundred times, with each time being randomized. Responses were then compared with the platform’s “default answers” to the same set of questions, which enabled the researchers to assess ChatGPT’s political stance.
“Although ChatGPT assures that it is impartial, the literature suggests that LLM [Large Language Model]s exhibit bias involving race, gender, religion, and political orientation,” the study states.
The research “can support the crucial duty of ensuring such systems are impartial and unbiased, mitigating potential negative political and electoral effects, and safeguarding general public trust in this technology,” the study concludes, highlighting the unique difficulty in detecting political bias, compared to gender or race.
The study’s lead author, Dr. Fabio Motoki, argued after publication, “Any bias in a platform like this is a concern,” regardless of whether the preference swings to the political left or right.
