For me or against me? Reactions to AI (vs. human) decisions that are favorable or unfavorable to the self and the role of fairness perception
Abstract
Public reactions to algorithmic decisions often diverge. While high-profile media coverage suggests that the use of AI in organizational decision-making is viewed as unfair and received negatively, recent survey results suggest that such use of AI is perceived as fair and received positively. Drawing on fairness heuristic theory, the current research reconciles this apparent contradiction by examining the roles of decision outcome and fairness perception on individuals' attitudinal (Studies 1--3, 5) and behavioral (Study 4) reactions to algorithmic (vs. human) decisions. Results from six experiments (N = 2,794) showed that when the decision was unfavorable, AI was perceived as fairer than human, leading to a less negative reaction. This heightened fairness perception toward AI is shaped by its perceived unemotionality. Furthermore, reminders about the potential biases of AI in decision-making attenuate the differential fairness perception between AI and human. Theoretical and practical implications of the findings are discussed.
Study specs
- Discipline
- Social Science,Artificial Intelligence,Psychology
- Year
- 2022
- Human Data Platform
- Prolific
- Source
- View Source DOI Google Scholar
Peer Review & Critical Discussion
Potential Selection Bias in 2023 Cohort
The participant pool shows a concerning overrepresentation of users from high-income demographics. Looking at Table 3, we can see that 78% of respondents had annual incomes above $75k, which significantly limits the generalizability of these findings to broader populations.
Non-naive Participants Issue
I've noticed a methodological concern regarding participant naivety. Given that Prolific users often complete multiple studies, there's a real risk that participants had prior exposure to similar experimental paradigms, which could confound the results.
RLHF Applicability to This Study Design
The implications for RLHF training pipelines are understated. If we accept the authors' conclusions about preference stability, this has direct consequences for how we should structure reward model training. The temporal decay effect described in Section 4.2 is particularly relevant.
Verify your expertise to join discussion
Create an account and verify your credentials to participate in peer discussions.