Opinion diversity through hybrid intelligence

This dissertation explores how Large Language Models (LLMs) can effectively and responsibly contribute to complex decision-making processes. By combining AI and human intelligence, Hybrid Intelligence (HI) emerges, allowing the strengths of both humans and machines to be utilized. The work focuses on applying HI in deliberations, a venue where citizens collectively discuss complex societal issues. The goal of these deliberations is not to make decisions but to understand different perspectives and encourage a diversity of opinions. While these gatherings often take place in person, online deliberations offer the opportunity to involve more people and increase diversity. At the same time, this presents a challenge: the vast amount of generated data makes insightful analysis difficult. Although AI can assist with automatic analyses, it struggles to identify viewpoints that are expressed infrequently. When humans and AI work together, on the other hand, we can uncover new insights and better highlight nuances in opinions. This shifts the focus from only obtaining high-level attitudes (for or against) to being able to capture the underlying arguments and values of the opinion holders. In HI, collaboration between citizens and AI is a process that is both transparent and reliable, which supports a democratic debate.<br>