AI in Debt Collection: Estimating the Psychological Impact on Consumers

The present study investigates the psychological and behavioral implications of integrating AI into debt collection practices using data from eleven European countries. Drawing on a large-scale experimental design (n = 3514) comparing human versus AI-mediated communication, we examine effects on consumers’ social preferences (fairness, trust, reciprocity, efficiency) and social emotions (stigma, empathy). Participants perceive human interactions as more fair and more likely to elicit reciprocity, while AI-mediated communication is viewed as more efficient; no differences emerge in trust. Human contact elicits greater empathy, but also stronger feelings of stigma. Exploratory analyses reveal notable variation between gender, age groups, and cultural contexts. In general, the findings suggest that AI-mediated communication can improve efficiency and reduce stigma without diminishing trust, but should be used carefully in situations that require high empathy or increased sensitivity to fairness. The study advances our understanding of how AI influences the psychological dynamics in sensitive financial interactions and informs the design of communication strategies that balance technological effectiveness with interpersonal awareness.

💡 Research Summary

The paper investigates how the use of artificial intelligence (AI) in debt‑collection communications influences consumers’ psychological and behavioral responses. Drawing on a large‑scale between‑subjects experiment conducted across eleven European countries, the authors recruited 3,514 adult participants (average age 35, 48 % female) via online crowdsourcing platforms. Participants were randomly assigned to read a scripted telephone interaction either with a human assistant (n = 1,775) or with an AI‑mediated assistant (n = 1,739). The scripts were identical in content—same debt amount, same repayment offer, same resolution path—but differed in interactional cues: the human script included longer waiting times and more pauses, whereas the AI script was portrayed as instantaneous and highly efficient.

After reading the script, participants rated six outcome variables on 5‑point Likert scales: perceived fairness, trust in the information, willingness to reciprocate (e.g., leave a review), perceived efficiency, feeling of stigmatization, and perceived empathy. Demographic data, as well as a brief Big‑Five personality inventory, were also collected. The authors tested five preregistered hypotheses: (H1) AI communication would be judged less fair, (H2) AI would be trusted less, (H3) AI would elicit less reciprocal behavior, (H4) human interaction would generate higher stigma, and (H5) human interaction would be rated as more empathetic.

Statistical analysis employed ordered logistic regression for each outcome, with assistant type as the primary predictor, controlling for age, gender, and country fixed effects. Interaction terms examined whether age or gender moderated the main effect. Robust standard errors were used; significance was set at α = 0.05.

Key findings:

• Fairness – participants rated human interactions slightly higher (M = 4.33, SD = 0.78) than AI (M = 4.24, SD = 0.77); the difference was statistically significant, supporting H1.
• Trust – no difference emerged; both conditions averaged 4.23 (SD ≈ 0.90), leading to a rejection of H2.
• Reciprocity – marginally higher intentions after human contact (M = 3.91) versus AI (M = 3.85), but the effect was not statistically significant, so H3 was not supported.
• Efficiency – AI‑mediated messages were rated as more efficient, aligning with the authors’ expectations.
• Stigma – participants felt more judged when the interlocutor was human, confirming H4.
• Empathy – human scripts generated higher empathy scores; the effect size (Cohen’s d = 0.34) indicates a modest but meaningful advantage, supporting H5.

Exploratory analyses revealed gender and age nuances: women tended to rate human assistants higher on fairness and empathy, while younger respondents showed a stronger preference for AI efficiency. Country‑level patterns suggested Scandinavian participants were especially favorable toward AI efficiency, whereas Southern European respondents placed greater emphasis on human fairness.

The authors interpret these results as evidence that AI can improve operational efficiency and reduce the emotional burden (stigma) associated with debt collection, without compromising trust. However, AI falls short on dimensions that rely on perceived moral agency and emotional resonance—fairness and empathy—highlighting the need for a hybrid approach where AI handles routine, efficiency‑driven tasks, and human agents intervene in contexts requiring nuanced moral judgment or emotional support.

Methodological limitations are acknowledged: the experiment used text‑based scripts rather than real voice calls, potentially limiting ecological validity; participants were generally debt‑naïve, which may not capture the heightened emotions of actual debtors; socioeconomic variables (education, income) were not controlled; and the relatively small number of countries prevented clustering of standard errors at the country level, possibly inflating precision. Moreover, the analysis omitted confidence intervals and model fit statistics, leaving some uncertainty about the robustness of the reported effects.

Future research directions include (1) field experiments with live phone calls to capture vocal tone and real‑time interaction dynamics, (2) manipulation of AI’s emotional expressiveness (e.g., tone, empathy cues) to test whether more “human‑like” AI can close the empathy gap, and (3) deeper cross‑cultural investigations that model cultural dimensions (e.g., power distance, uncertainty avoidance) as moderators.

In conclusion, the study provides the first large‑scale empirical evidence that AI‑mediated debt‑collection communication can enhance efficiency and lessen stigma, yet it cannot fully replace the human capacity for perceived fairness and empathy. Practitioners are advised to adopt a balanced strategy that leverages AI for routine efficiency while retaining human involvement for emotionally sensitive negotiations.