Ruben Durante grew up in Sicily, Italy and obtained his doctorate from Brown University in 2010. He started his career as an Assistant Professor of Economics at Sciences Po in Paris and spent several years at Universitat Pompeu Fabra in Barcelona before joining NUS in 2023. He currently holds the Provosts’ Chair at the NUS Department of Economics. Ruben studies political economy with a focus on the functioning and impact of traditional and new media in mature and consolidating democracies. He is also interested in questions related to identity, culture, and redistribution. His work has been published in top journals in economics, political science, and management, and is regularly featured in the press. His research has been supported by several funding agencies including the prestigious European Research Council (ERC).

Have you always wanted to become an economist?

I did not set out to be an academic economist—despite coming from a family of academics. However, as a teenager, I came across a book that completely changed my perspective: Banker to the Poor by Muhammad Yunus. Fascinated by the promise of microcredit, I wrote my undergraduate thesis on a South American microcredit experiment and imagined a career in development economics. Over time, my interests shifted towards political economy and the role of media, the internet, and lately AI, which is what my research has mainly been about.

Please tell us about your recent research on political economy, the media and AI.

I study how information flows shape accountability—of politicians, public institutions, and powerful private interests—across both democratic and authoritarian settings. Let me discuss two recent projects, which probe timely questions: how AI reshapes trust in the media, and how authoritarian regimes reengineer information online.

AI and the Paradox of Trust in the Media

The first paper, “Generative AI and Trust in the Media: Evidence from a Large-scale Field Experiment”, co-authored with Filipe Campante, Ananya Sen, and Felix Hagemeister, is based on an experiment we ran with subscribers of Süddeutsche Zeitung (SZ), a German newspaper.

The goal of the paper is to test how people react when they realize they cannot distinguish real from AI-generated content anymore. Misinformation is a big problem nowadays, and there is a fear that AI could lead to a complete collapse of the public’s trust in all media and institutions more generally.

Working directly with SZ’s subscriber base, we ran a field experiment. A treatment group was shown three pairs of images—one real and one AI-generated in each pair—and asked to identify which was which (including the options “both” or “neither”). The images were intentionally convincing. The idea was to prime participants to confront the new reality that their ability to detect fakes may be limited.

We then measured two kinds of outcomes. First, survey measures captured shifts in beliefs and trust in media. Second, we used SZ’s first-hand data to observe actual news engagement, subscription retention, and reading behaviour in the days and weeks after the intervention.

What we found is interesting and unexpected. On the one hand, the treatment reduced trust in “the media” broadly construed, including in SZ itself. However, engagement with the newspaper actually rose. Despite lower stated trust, treated subscribers increased their consumption of SZ content in the days following the experiment and were more likely to remain subscribed in subsequent weeks and months.

We have a simple theoretical model that helps reconcile the paradox. When trust becomes scarce, trustworthiness becomes more valuable. Even if readers revise their beliefs downward about an outlet’s absolute trustworthiness, high-reputation sources remain the best available option in a worse information environment. For high-quality news organisations, this creates both a responsibility and an opportunity: to help audiences navigate uncertainty and authenticate content in an AI-saturated environment.

How Autocrats Rewrite the Internet

The second paper, “The Anatomy of Censorship and Propaganda: Evidence from Russian Wikipedias”, co-authored with Vladimir Avetian, Ulrich Matter, and Katia Zhuravskaya, examines online censorship and propaganda by studying online encyclopaedias in Russia.

In Russia, information is heavily controlled by the government, so a lot of people have relied on Wikipedia to gather reliable information. For years, Russian authorities struggled to bend Wikipedia to their preferred narrative. Around mid-2023, their strategy changed: pro-Kremlin backers cloned Wikipedia’s Russian-language content onto domestic servers, then spent seven months silently editing it. In January 2024 they launched a native competitor: RuWiki.

Because every change on Wikipedia is versioned, we can document every single change that was implemented across over 3 million pages. This allows us to get into the autocrat’s mind: if he could perfectly choose what people could and could not read, what would be his priorities?

We find three main patterns. First, we find that RuWiki’s editing is centralised and professional. Compared to Wikipedia, there are far fewer contributors, who follow a 9-to-5, Monday-to-Friday schedule. This is consistent with paid editorial work rather than volunteer effort. Second, content that is hard to reframe is deleted completely. For example, about 2,000 pages on topics such as political rights in Russia and the war in Ukraine, and content framed as violating “traditional values”, such as sex, pornography, and LGBTQ topics, were deleted completely. Third, content on the surviving pages was systematically reframed. Using text-comparison tools akin to anti-plagiarism software, we tracked which terms and phrases were added and which disappeared. References like “war” or “invasion of Ukraine” were replaced with “special military operation,” and articles were rewritten to promote “traditional values” and the image of political elites.

To summarize, the edits look like a professional effort to legitimise the current regime, recast the Ukraine invasion, and position Russia as the guardian of tradition against a morally declining West.

What do you think is the role of economists in an AI-filled future?

I am quite optimistic. Our edge lies in rigorous thinking about complex social problems, and I believe this will become more important in the future, not less. Even though AI systems are becoming better and better, humans will maintain a comparative advantage in identifying deep questions, causal structures, and designing creative settings that separate correlation from causation. These skills actually become more valuable as automated tools get better.

AI should be embraced as a complement: for data collection, organisation, and even parts of analysis and writing that are procedural rather than conceptual. The challenge for universities is to teach students how to wield these tools efficiently without outsourcing the core human tasks I just mentioned.

If you want to learn more about this, take my courses!

March, 2026