Course Catalog
AI and Democracy
94-816
Units: 6
Description
AI is already part of democratic life. Voters ask AI chatbots who should get their vote, campaigns use AI to write ads, and foreign influence operations use it to run social influence campaigns. Governments are starting to use it to analyze public comments, and authoritarian governments use it to extend surveillance and censorship. A healthy information ecosystem is how a society notices problems and decides how to respond, including to AI’s effects on human flourishing. This course asks whether AI will strengthen or weaken that capacity.
There is also early evidence that AI can help. In a 2024 Science study, conversations with an AI chatbot reduced belief in conspiracy theories by about 20 percent, and an independent team reproduced the effect in 2026. In another 2024 Science study, participants preferred group statements written by an AI mediator to those written by human mediators, and many moved toward a shared view. Whether AI strengthens or weakens democracy depends on how we shape its development and use.
The course starts with democratic theory. Robert Dahl argued that a strong democratic process should give every adult citizen an equal chance to learn about the issues, set the agenda, make their views known, and vote. Students will learn democratic values and use them to judge how AI tools respond, and consider whether AI concentrates power in politics, government, and news, or spreads it more widely.
Much of class time is hands-on. Students run a deliberative poll on a local issue and test AI models from different countries against political questions in different languages. They make short deepfakes of themselves and test whether classmates and chatbots can spot them. Students run a crisis exercise based on a real foreign influence operation. They play skeptical voters to test how chatbots respond to election conspiracy claims.
The final project combines policy and technical skills. Groups of 2-4 either build a working prototype, such as a tool that flags AI-generated public comments, or build a model evaluation that tests how AI systems handle a democratic task, such as whether chatbots tell voters what they want to hear or answer political questions differently in Chinese and English. Groups present their work in a demo and a presentation.
Learning Outcomes
Students will be able to:
- Explain core ideas of democratic and constitutional theory.
- Describe how AI is changing elections and public debate, and assess future scenarios, including one in which citizens act through AI agents.
- Assess whether AI concentrates or disperses political, governmental, and informational power, and compare how democratic and authoritarian governments use it.
- Audit how leading AI models answer contested political questions, and explain where their political values come from.
- Analyze how AI spreads misinformation and disinformation, and weigh the evidence that it can also correct false beliefs.
- Evaluate AI tools for public listening and deliberation, including whether they support citizens’ own judgment.
- Build or evaluate an AI application for a democratic problem in a small group that combines policy and technical skills.
Prerequisites Description
None.