British scientists found that ChatGPT adheres to the left
[ad_1]
Scientists from the University of East Anglia found that ChatGPT adheres to left-wing political views, reports Sky News. In response to questions from researchers, the popular OpenAI neural network rated the activities of left-wing political forces in Britain, the United States and Brazil much more positively than their right-wing opposition.
As part of the study, scientists made ChatGPT about 100 times to complete a survey consisting of more than 60 statements on political topics. At the same time, the neural network was first asked to evaluate the statements, imagining that it was a representative of each of the political movements – from the far right to the extreme left – and then asked it to evaluate these same statements from a neutral point of view. Comparing the neural network’s ideas about neutrality with the answers on behalf of representatives of political movements, the scientists found out that the most neutral position of ChatGPT coincides with its own “left” answers.
“Sometimes people forget that these AI models are just machines. They give a very plausible, overcooked summary of what you’re asking, even if it’s fundamentally wrong. And if you ask her: “Are you neutral?”, She will answer: “Oh, yes, I am neutral!”, – one of the authors of the study, Dr. Fabio Motoki, told Sky News. The study says that the results showed a clear commitment to the views of the neural network “democrats in the United States, Lula in Brazil (President Luiz Inacio Lula da Silva.— “b”) and the Labor Party in Britain”. At the same time, as Dr. Motoki noted, if the neural network supported right-wing politicians, there would be just as many reasons for concern. “Just like the media, the internet and social media, it can influence the public and that can be very harmful.”
[ad_2]
Source link
