OpenAI’s leading chatbot technology, ChatGPT, is reportedly displaying a staggering leftist tilt, recent research suggests. A team of academics at the University of East Anglia in England led a comprehensive probe into the AI system’s political leanings, revealing a notable systemic bias towards leftist ideologies and political organizations, including the Democratic Party in the United States, Brazil’s ex-president Lula, and the UK’s Labour Party.
The researchers undertook a well-defined study, asking politically charged questions of ChatGPT. They asked the AI to answer in the manners of a Republican, a Democrat, and of neutral alignment respectively. By juxtaposing and analyzing the responses, they noted significant discrepancies that notably favored left-wing perspectives.
This isn’t the first time an alarm has been raised over OpenAI’s ChatGPT’s potential bias. Earlier this year, a considerable evaluation released by the Manhattan Institute, unveiled that ChatGPT has an alarming tolerance towards maligning conservatives than liberals, when posed with comparable remarks.
The UK-based researchers took their experiment further to solidify their assertions. They went ahead to repeat the same politically-charged questions a 100 times, then ran each response through “1,000 repetitions for each answer and impersonation.”
This was performed to constrain the chatbot’s inherent randomness and inclination to propagate misleading or incorrect information. The scientific undertaking uncovered “grave worries” that the AI systems, including ChatGPT, can potentially escalate or amplify the pressing challenges threatening political discourse on the internet and social media.
