Study Finds ChatGPT Displays Left-Wing Bias in Responses

Researchers warn that generative AI's political leanings could influence public discourse and undermine democratic trust.

A study published in the Journal of Economic Behavior & Organization found that ChatGPT's responses align more closely with left-wing perspectives compared to average American opinions.
Researchers used text and image analysis to assess political bias, revealing systematic deviations in ChatGPT's outputs toward left-leaning views, with occasional exceptions on topics like military supremacy.
ChatGPT was observed to decline generating content aligned with right-wing perspectives on certain themes, citing misinformation concerns, while producing left-leaning content without hesitation.
Experts highlight the potential societal risks of unchecked biases in generative AI, such as deepening divides, eroding trust in institutions, and distorting democratic processes.
The study calls for interdisciplinary collaboration among policymakers, technologists, and academics to create AI systems that are fair, accountable, and aligned with societal norms.