AI & Democracy
Academic course (2h)
A two-hour session exploring how LLMs interact with deliberation, participation, and large-scale decision processes in a democratic context.
AI, Democracy & Collective Intelligence
TL;DR
- A practical framework to understand how LLMs influence deliberation and collective decision-making.
- Two available formats: academic course (2h) and technical talk (45 min).
- Open resources: bibliography, notes, and slides.
How can large language models support - rather than undermine - deliberation, participation and collective decision-making? This page documents my ongoing exploration of how AI is changing decision-making processes in organisations and in democratic life, whether at the local or global scale.
Why this project?
I work in applied AI and have spent several years building real-world systems. In parallel, I am increasingly concerned by the impact of AI on democratic processes: who speaks, who is heard, how decisions are made, and what role automated systems play in all of this. My goal is simple: understand how we can use new language models (LLMs) to strengthen democratic practices instead of weakening them.
I approach this topic from three angles:
- Citizen (40%) — understanding and contributing to issues that affect us collectively.
- Intellectual (40%) — reading, structuring ideas, and connecting disciplines.
- Professional (20%) — reflecting on the future of AI practice and applied ethics.
Explore my work
This section presents my work on AI's impact on decision-making, including two complementary approaches along with teaching materials, curated references, and research notes.
Can LLMs improve collective decision-making?
What research tells us
Tech talk (45 min)
Format : Presentation followed by Q&A
When a group needs to make a decision (in a team, workshop, or organization), we often expect the collective to perform better than its individual members. However, the arrival of language models (LLMs) in these contexts...
References & reading list
The knowledge base behind this project. Includes sources cited in the lecture and my active reading list on the topic.
Explore the library →Notes, ideas & open questions
I am particularly interested in questions such as:
- When do LLMs help people understand each other better - and when do they distort?
- How can we design deliberative processes that remain legitimate when AI is involved?
- What does "democratic alignment" of models mean in practice, beyond slogans?
As this project grows, I will publish short notes and reflections here.
My approach
1. Read and map the field
I conduct an exploratory but in-depth review of recent work on AI-mediated deliberation, collective intelligence, and democratic innovation - including systems like Polis, experiments in "AI-mediated deliberation", and the intersection of computer science and political theory.
2. Prototype and experiment
Theory is not enough. I apply my engineering background to build prototypes and Proofs of Concept (POCs). Testing how LLMs interact with real political arguments allows me to confront academic promises with technical reality (latency, hallucinations, bias).
3. Teach and share
I synthesized this research into a comprehensive module on AI & collective intelligence. It bridges the gap between technical constraints (transformers, bias) and democratic theory to explore concrete applications for organizations. This website gradually becomes the place where I publish notes, slides, and document my ongoing thinking.
4. Connect with the field
Democracy is a collective effort. I actively engage with researchers, civic tech builders, and institutions to ground these ideas in practice and avoid the "techno-solutionism" trap.