|
on Nudge and Boosting |
|
Issue of 2026–09–21
four papers chosen by Marco Novarese, Università degli Studi del Piemonte Orientale |
| By: | Haya Halimeh; Sascha Kaltenpoth; Kevin B\"osch; Oliver M\"uller |
| Abstract: | LLM-based GUI agents increasingly act on behalf of users in digital environments that were designed with human users in mind. These graphical user interfaces were designed to support, but also deliberately steer, the behaviour and decisions of users. While behavioural biases in the textual outputs of LLMs are well-documented, far less is known about how such influence operates when models act as agents that perceive interfaces and execute decisions---and, in particular, whether the reasoning capabilities increasingly built into these agents make them more robust to it. Drawing on Dual-Process Theory, we empirically investigate whether LLM-based GUI agents are susceptible to automatic (Type 1) and reflective (Type 2) digital nudges, and how their reasoning configuration moderates this susceptibility. In a randomized online shopping experiment with 3, 600 agents and a total of 21, 600 simulations across six frontier models from three providers, we found that agents were vulnerable to both nudge types. Crucially, the reasoning configuration moderated these effects in opposing directions, reducing susceptibility to automatic default nudges while heightening it to reflective social influence nudges. Extensive reasoning therefore did not make agents more robust but redirected the route through which choice architecture takes effect. Exploratory analysis further showed this redirection to be systematically structured by model scale. Beyond establishing nudge susceptibility as a behavioural property of agentic AI, the study positions interface design as a governance concern for organizations that delegate decisions to autonomous agents. |
| Date: | 2026–09 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2609.19843 |
| By: | Xiaodan Shao; Vivek Choudhary; Arnab Majumdar |
| Abstract: | Drug-drug interaction (DDI) errors pose serious risks to patient safety. Existing decision-support systems often require physicians to respond to alerts, disrupting workflows and contributing to high override rates. We examine whether a non-mandatory information intervention can reduce DDI errors and foster learning. Using a randomized field experiment with India's largest electronic medical record platform, we analyze 2.81 million prescriptions from 1, 700 physicians using a difference-in-differences design. Treatment physicians received real-time information highlighting DDI errors without being required to respond, while control physicians received no such information. The intervention reduced DDI errors by 8.6%, corresponding to an estimated US$4.8 million in annual hospitalization cost savings and approximately 134 lives potentially saved. We identify two mechanisms: reactive correction, whereby physicians remove errors after they are flagged, and proactive learning, whereby they avoid errors before alerts occur. While early reductions are driven primarily by correction, physicians increasingly avoid errors over time. They also become less likely to repeat previously flagged errors and reduce new errors, suggesting that learning generalizes beyond specific drug pairs. The effects are consistent across physician types and do not compromise productivity or care quality. Our findings show that non-mandatory information interventions can improve patient safety through both immediate error correction and persistent, generalizable learning. |
| Date: | 2026–09 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2609.09673 |
| By: | Evangelos Dioikitopoulos; Rigissa Megalokonomou; Tommaso Sartori; Yves Zenou |
| Abstract: | Public information campaigns often rely on credible expert messengers to influence people’s beliefs. We study how messenger identity and communication technology shape persuasion in a large-scale randomized experiment on vaccination attitudes among young adults in Greece. Participants receive an identical public-health message that varies only in who delivers it — a medical scientist or a social media influencer — and in how it is delivered — through immersive virtual reality or conventional video. Exposure raises trust in vaccines and trust in the healthcare system. Scientists are rated substantially more credible than influencers yet are no more persuasive on average; the two groups’ effectiveness rests on different characteristics — the scientists’ on perceived reliability and knowledgeability, the influencers’ on personal appeal, prior familiarity, and reputation. We find that persuasion is greater under virtual reality, consistent with stronger engagement. Changes in beliefs persist for at least six months, and initially hesitant participants report greater willingness to vaccinate. A model in which engagement, prior uncertainty, and effective credibility enter persuasion multiplicatively rationalizes these patterns. |
| Keywords: | randomized experiment, persuasion, belief formation, source credibility, messenger familiarity, social media influencers, virtual reality, engagement, health communication, vaccination |
| JEL: | C93 D83 I12 D91 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_12961 |
| By: | Linda Dezso (University of Graz, Austria; EcoAustria - Institute for Economic Research, Austria); Christian Koch (University of Aberdeen, United Kingdom) |
| Abstract: | Public demands to restrict immigrants' access to welfare typically invoke immigrants' outgroup status and their allegedly insufficient contributions. While contributions are fundamental to the welfare state, immigrants' contributions are not directly observable, which creates fertile ground for misperceptions that political rhetoric often exploits when portraying natives as the main contributors. We present results from a stylized experiment in which UK-resident natives and immigrants are paired to divide resources. We isolate how ingroup favoritism and framed contributor status shape distributive choices and fair-share beliefs, even in the absence of actual differences in contributions to the resources. We find no statistically significant effect of ingroup favoritism on choices, but both natives and immigrants claim more when framed as contributors. Making unequal allocations costly reduces claims, and secondary analyses suggest that costliness also attenuates the contributor effect. Both groups also report higher fair-share beliefs when framed as contributors than when they are not. These findings show that merely framing individuals as contributors shapes distributive preferences in self-serving ways and produces egocentrically biased fair-share beliefs. We propose that incorporating such self-serving dynamics into the microfoundations of preferences for redistribution toward immigrants could help explain exclusionary welfare preferences beyond explanations centered on anti-immigrant attitudes and nationalism. |
| Keywords: | contributions, framing, redistribution, ingroup favoritism, self-serving behavior, fairness beliefs, natives, immigrants, spectators, UK |
| JEL: | C9 D63 D69 D91 J15 |
| Date: | 2026–09 |
| URL: | https://d.repec.org/n?u=RePEc:grz:wpaper:2026-18 |