Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The Data-Dollars Tradeoff: Privacy Harms vs. Economic Risk in Personalized AI Adoption

Alexander Erlei, Tahir Abbas, Kilian Bizer, Ujwal Gadiraju

TL;DR

It is found that ambiguity over data leaks, rather than only privacy preferences per se, drives avoidance behavior among users towards personalized AI, suggesting strong demand for transparency institutions.

Abstract

Privacy concerns significantly impact AI adoption, yet little is known about how information environments shape user responses to data leak threats. We conducted a 2 x 3 between-subjects experiment (N=610) examining how risk versus ambiguity about privacy leaks affects the adoption of AI personalization. Participants chose between standard and AI-personalized product baskets, with personalization requiring data sharing that could leak to pricing algorithms. Under risk (30% leak probability), we found no difference in AI adoption between privacy-threatening and neutral conditions (ca. 50% adoption). Under ambiguity (10-50% range), privacy threats significantly reduced adoption compared to neutral conditions. This effect holds for sensitive demographic data as well as anonymized preference data. Users systematically over-bid for privacy disclosure labels, suggesting strong demand for transparency institutions. Notably, privacy leak threats did not affect subsequent bargaining behavior with algorithms. Our findings indicate that ambiguity over data leaks, rather than only privacy preferences per se, drives avoidance behavior among users towards personalized AI.

The Data-Dollars Tradeoff: Privacy Harms vs. Economic Risk in Personalized AI Adoption

TL;DR

It is found that ambiguity over data leaks, rather than only privacy preferences per se, drives avoidance behavior among users towards personalized AI, suggesting strong demand for transparency institutions.

Abstract

Privacy concerns significantly impact AI adoption, yet little is known about how information environments shape user responses to data leak threats. We conducted a 2 x 3 between-subjects experiment (N=610) examining how risk versus ambiguity about privacy leaks affects the adoption of AI personalization. Participants chose between standard and AI-personalized product baskets, with personalization requiring data sharing that could leak to pricing algorithms. Under risk (30% leak probability), we found no difference in AI adoption between privacy-threatening and neutral conditions (ca. 50% adoption). Under ambiguity (10-50% range), privacy threats significantly reduced adoption compared to neutral conditions. This effect holds for sensitive demographic data as well as anonymized preference data. Users systematically over-bid for privacy disclosure labels, suggesting strong demand for transparency institutions. Notably, privacy leak threats did not affect subsequent bargaining behavior with algorithms. Our findings indicate that ambiguity over data leaks, rather than only privacy preferences per se, drives avoidance behavior among users towards personalized AI.
Paper Structure (11 sections, 2 equations, 11 figures, 4 tables)

This paper contains 11 sections, 2 equations, 11 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Experimental Design
  4. Procedure: A Detailed Overview
  5. Outcomes of Interest
  6. Predictions
  7. Results
  8. Discussion
  9. Implications
  10. Caveats, Limitations, and Future Work
  11. Conclusions

Figures (11)

  • Figure 1: Illustration of the experimental flow, and the four stages that participants completed in our study.
  • Figure 2: The proportion of subjects consenting to share their data with the AI personalization system.
  • Figure 3: Top: Average feelings of betrayal as measured by the post-experimental questionnaire. Following kormylo2025till, we use an average of three questions that elicit feelings of violated trust, feelings of betrayal, and feeling like one made a mistake, after observing the data leak (lottery). Bottom: Average feelings of betrayal across treatments and the AI personalization choice.
  • Figure 4: Rejection rates in the framed ultimatum game.
  • Figure 5: Feelings of (1) betrayal, (2) trust violations and (3) having made a mistake (from left to right) according to self-reported measures in the post-experiment questionnaire.
  • ...and 6 more figures