Accidental Bias or a Finger on the Scale?
September 18, 2023
Who knew? According to Bezos’ rag The Washington Post, “Chat GPT Leans Liberal, Research Shows.” Writer Gerrit De Vynck cites a study on OpenAI’s ChatGPT from researchers at the University of East Anglia:
“The results showed a ‘significant and systematic political bias toward the Democrats in the U.S., Lula in Brazil, and the Labour Party in the U.K.,’ the researchers wrote, referring to Luiz Inácio Lula da Silva, Brazil’s leftist president.”
Then there’s research from Carnegie Mellon’s Chan Park. That study found Facebook’s LLaMA, trained on older Internet data, and Google’s BERT, trained on books, supplied right-leaning or even authoritarian answers. But Chat GPT-4, trained on the most up-to-date Internet content, is more economically and socially liberal. Why might the younger algorithm, much like younger voters, skew left? There’s one more juicy little detail. We learn:
“Researchers have pointed to the extensive amount of human feedback OpenAI’s bots have gotten compared to their rivals as one of the reasons they surprised so many people with their ability to answer complex questions while avoiding veering into racist or sexist hate speech, as previous chatbots often did. Rewarding the bot during training for giving answers that did not include hate speech, could also be pushing the bot toward giving more liberal answers on social issues, Park said.”
Not exactly a point in conservatives’ favor, we think. Near the bottom, the article concedes this caveat:
“The papers have some inherent shortcomings. Political beliefs are subjective, and ideas about what is liberal or conservative might change depending on the country. Both the University of East Anglia paper and the one from Park’s team that suggested ChatGPT had a liberal bias used questions from the Political Compass, a survey that has been criticized for years as reducing complex ideas to a simple four-quadrant grid.”
Read more about the Political Compass here and here. So does ChatGPT lean left or not? Hard to say from the available studies. But will researchers ever be able to pin down the rapidly evolving AI?
Cynthia Murrell, September 18, 2023
Comments
-
- Subscribe to Beyond Search
Feature archive
News archive
-
