Reaching consensus in a democracy can be challenging due to varied ideological, political, and social views. Researchers at Google DeepMind, however, have developed an AI tool that could help bridge such divides.
Their system, known as the Habermas machine (HM), is designed to act as a “caucus mediator” by generating summaries that highlight areas of agreement among people discussing complex social or political issues.
The AI model is not meant to persuade but to mediate, says Michael Henry Tessler, a research scientist at Google DeepMind. The tool’s purpose is to identify common ground by analyzing input from group members. The study, published in Science, recruited over 5,700 participants and tested the system’s ability to summarize group opinions and facilitate discussions.
The system relies on two language models. One generates statements that reflect the group's diverse views, while the other evaluates how likely participants are to agree with these statements. The AI-generated summaries were then compared with those produced by human mediators.
In a series of experiments, participants often preferred the AI’s summaries over those of the human mediators. About 56% of the time, they found the AI-generated statements to be of higher quality and endorsed them more strongly. After deliberations, participants also showed a reduction in divisiveness.
Despite its success, the researchers caution that the system has limitations. It does not fact-check, keep discussions on topic, or moderate discourse effectively. Joongi Shin, a researcher at Aalto University, also warns that ethical issues may arise if users are unaware of the AI’s role in the process. Further research is needed to responsibly deploy such technology in real-world deliberations.
.png)
.png)
0 Comments