CEPS AI

Artificial Intelligence, Governance, and Society

CEPS AI is a research initiative devoted to the study of the economic, institutional, and societal implications of artificial intelligence and algorithmic systems. Building on the strengths of CEPS in economic theory and public policy, the initiative examines how AI transforms markets, public decision-making, democratic processes, and the circulation of information.

CEPS AI focuses on how artificial intelligence reshapes economic interactions, information environments, and collective decision-making in modern economies and societies. Particular attention is paid to governance, regulation, and the design of institutions and public policies that ensure responsible, transparent, and socially beneficial uses of algorithmic systems.

By bringing together researchers at the intersection of economics, public policy, data science, and political science, CEPS AI provides a platform for interdisciplinary research and dialogue on the challenges raised by AI in contemporary societies.

At CEPS, Artificial Intelligence Serves as a Lens for Economic and Social Analysis

AI & Democratic Information

Thomas Renault — Professeur at RITM and CEPS — www.thomas-renault.com

How does the use of large language models in the public sphere transform the production, dissemination, and evaluation of information, and with what consequences for democratic debate?

Thomas Renault studies the use of large language models (LLMs) on social media—more specifically on X, to analyze the growing role of conversational AI systems as information intermediaries. Drawing on original empirical data, this work examines how users employ LLMs for fact-checking, the types of questions they pose, who initiates these fact-checks, and who is being fact-checked. It also analyzes the degree of agreement between different LLMs, between LLMs and evaluations from the crowd-based Community Notes program, and between LLMs and professional fact-checkers. The findings highlight key challenges for democracy related to trust, transparency, and the concentration of informational power in the hands of a small number of technological actors, contributing to broader reflections on AI governance and the institutional conditions required to preserve a pluralistic and reliable information environment.

References

Renault, T., Mosleh, M., & Rand, D. G. (2025). @Grok Is This True? LLM-Powered Fact-Checking on Social Media. Working Paper.
Renault, T., Mosleh, M., & Rand, D. G. (2025). Republicans are flagged more often than Democrats for sharing misinformation on X’s Community Notes. PNAS, 122(25), e2502053122.

AI & Financial Sustainability

Samuel Ligonnière — Assistant Professeur at CEPS.

How can artificial intelligence help assess the sustainability and resilience of firms and financial systems in the face of environmental and economic risks?

Work led by Samuel Ligonnière examines how AI can aggregate climate-related information disclosed by firms, assess its reliability, and identify inconsistencies across sources, as well as how environmental news is priced in financial markets. This research sheds light on the mechanisms underlying green finance and investor responses to climate-related risks. Related work applies AI tools to the study of corporate distress and insolvency in France, analyzing the determinants of firm failures, the allocation of firms across restructuring and bankruptcy procedures, and their economic consequences across sectors and territories. Together, these projects contribute to a broader understanding of financial sustainability, market resilience, and the role of AI in informing economic policy.