The greatest challenges facing humanity—from global warming to the spread of misinformation—are problems we have made for ourselves; human problems that require human solutions. Our project starts from the idea that many of these challenges stem from failures of reasoning, especially a tendency toward closed-mindedness. In particular, we explore whether recent advances in AI, especially Large Language Models, can help people engage more thoughtfully with ideas that challenge their beliefs.
Bringing together insights from philosophy and psychology, we ask two central questions: ‘What does it mean to be open-minded?’ and ‘How can new technologies help us measure and encourage it?’ We propose that truly open-minded people are willing to seriously consider alternative viewpoints and revise their beliefs when the evidence warrants it—bringing them closer to accurate understanding.
To study this process, we use structured conversations between participants and AI systems. In these dialogues, the AI produces strong evidence-based arguments, allowing us to observe how and when belief change occurs. Across a range of topics where people often hold inaccurate beliefs, we test whether conversations with AI can help people reflect more carefully and update their views, ideally in a way that genuinely increases subsequent open-mindedness.
While using AI as a replacement for thinking—for example, to generate answers to homework—may undermine independent thought, our research explores how AI can instead support intellectual autonomy by helping people evaluate evidence more carefully. Ultimately, this project aims to deepen our understanding of open-mindedness and how it develops. By combining philosophical analysis, psychological research, and emerging AI technologies, we hope to design tools that encourage thoughtful reflection, constructive dialogue, and a greater willingness to revise beliefs when the evidence calls for it.
