Titled “More human than human: measuring the political leaning of ChatGPT,” a team of researchers from Brazil and the United Kingdom published last Wednesday (16) brought to light an intriguing study of ChatGPT’s political leanings. Artificial Intelligence platform. The study concluded that ChatGPT preferred the views of voters of the US Democratic Party, former President Luiz Inacio Lula da Silva (PT) and the UK Labor Party.
The researchers responsible for the analysis, Fabio Motogi, Waldemar Pinho Neto and Victor Rodriguez, developed a sophisticated approach to identify ideological biases in the responses ChatGPT provided to users. They proposed that ChatGPT responded with a certain ideological bias, and compared these responses to the site’s “default” responses without any political instruction.
According to the researchers, the study found “solid evidence” of significant and systematic political bias in favor of the Democrats in the US, President Lula in Brazil and the Labor Party in the UK. The researchers ran experiments asking the AI to adopt the position of a supporter of Lula’s former president, Jair Bolsonaro (PL), without any political profiling. Surprisingly, “neutral” responses also showed similarities with favorable responses to PT.
The tests used questions from the British-style “Political Compass,” which measures political will on economic and social issues. ChatGPT responses were consistently associated with the previously identified left bias.
The researchers cautioned that these findings have far-reaching implications, suggesting that text generation mechanisms such as ChatGPT may exacerbate existing policy challenges posed by the Internet and social media environment. They urge politicians, policy makers, the media and academics to consider these results carefully when dealing with technology and policy issues.
With information from Cointelegraph.
Did you know there is a Jurists Portal? Facebook, Twitter, Instagram, telegram, Share it, Google News This is Linkedin? Follow us
