(qlmbusinessnews.com Thur, 17th Aug, 2023) London, UK —
A recent study conducted by researchers at the University of East Anglia in the UK has unveiled a noteworthy left-wing bias within ChatGPT, a widely used artificial intelligence chatbot. The investigation highlighted the chatbot's tendency to display a preference for the UK's Labour Party and the Democratic Party led by President Joe Biden in the United States.
——————–Advertisement————————
Though concerns regarding potential political bias within ChatGPT have been previously raised, most notably by business magnate Elon Musk, this study is the first to provide comprehensive evidence of such favouritism. Lead author Dr Fabio Motoki expressed concern over the implications of these findings as the OpenAI platform gains popularity among the public, particularly in light of upcoming elections on both sides of the Atlantic.
Dr Motoki emphasized, “Any bias in a platform like this is a concern… Sometimes people forget these AI models are just machines. They provide very believable, digested summaries of what you are asking, even if they're completely wrong. And if you ask it ‘are you neutral,' it says ‘oh I am!' Just as the media, the internet, and social media can influence the public, this could be very harmful.”
Testing ChatGPT for Bias
To ascertain the extent of the chatbot's bias, researchers instructed ChatGPT to simulate responses from individuals representing various political perspectives while addressing a range of ideological questions. These queries spanned from extreme to neutral viewpoints, with each “individual” responding with agreement, strong agreement, disagreement, or strong disagreement.
The responses were then compared to ChatGPT's default answers for the same set of inquiries, allowing researchers to gauge the degree to which the responses aligned with specific political stances. Each of the 60-plus questions was posed 100 times to account for potential variations in the AI's output, and these multiple responses were further analyzed for indications of bias.
This methodology aimed to replicate the structure of a survey administered to a genuine human population, recognizing that answers may differ based on when they are provided.
Origins of the Bias
ChatGPT draws on an extensive dataset of text sourced from the internet and beyond. The researchers suggested that this dataset might inherently contain biases that influence the chatbot's responses. Additionally, the underlying algorithm used for training the AI could contribute to amplifying existing biases within the data it has absorbed.
——————–Advertisement————————
The research team plans to release their analysis approach as a free tool for individuals to assess potential biases within ChatGPT's output. Co-author Dr Pinho Neto stated, “We hope that our method will aid scrutiny and regulation of these rapidly developing technologies.”
The study's findings have been published in the journal Public Choice.
This News Story is brought to you by QLM Business News, your Digital Media Channel
Visit QLM businessnews.com
For more business news stories also follow us on Facebook, Twitter and Youtube.
To Help QLM Business News bring you more new stories like this, please like, share and subscribe.
