Columnist suggests AI could affect polarization in the US — Taipei Times
In the United States, large language models and chatbots may help soften political polarization because they generate responses based on the most likely conclusions from vast amounts of data. This assessment was made by Stephen Mihm, associate dean and professor of history at the University of Georgia, in a Bloomberg Opinion column published by the Taipei Times.
Mihm writes that more and more people receive information not through searches and link clicks, but from brief AI summaries or chatbot responses. In his view, such systems usually gravitate toward average, most widespread assessments and may offer users more moderate explanations of political topics.
From television networks to digital communities
The author links the intensification of polarization in the United States to changes in the media environment. In the 1950s and 1960s, radio and television provided mass audiences with largely the same information through a limited number of networks. At the same time, Mihm believes this model simplified reality and pushed dissenting voices to the periphery.
More current news is available on the UA.News Telegram channel Telegram.
With the spread of cable television, the number of channels rose sharply. After the Cable Act was passed in 1984, barriers to the emergence of cable networks declined: in 1998, the United States had 174 national cable networks, compared with 82 six years earlier, while thousands of local stations also operated. The internet and social media, Mihm writes, further intensified the division of audiences into digital communities with different information realities.
The risk of new groupthink
The columnist mentions a Brennan Center for Justice report on election disinformation. In his account, chatbots did not amplify marginal theories about election fraud, but instead refuted such claims. He also cites a study in the Journal of Computer-Mediated Communication, according to which people who discussed political issues with chatbots tended to hold more moderate views.
At the same time, Mihm cautions that AI has an inherent tendency toward generally accepted positions and may crowd out minority views or unconventional ideas. In his opinion, reducing polarization through broader AI use may come at the cost of new groupthink that is dangerous for democracy.
