[Paper Review] Overcoming Anchoring Bias: The Potential of AI and XAI-based Decision Support

Felix Haag, Carlo Stingl+3 more|arXiv (Cornell University)|May 8, 2024

Big Data and Business Intelligence4 citations

TL;DR

This study investigates whether AI and explainable AI (XAI)-based decision support can mitigate anchoring bias in purchase decisions. Using two online experiments with 390 participants, it demonstrates that both AI alone and AI combined with XAI significantly reduce the influence of anchoring, offering a pathway to design more ethical and unbiased information systems.

ABSTRACT

Information systems (IS) are frequently designed to leverage the negative effect of anchoring bias to influence individuals' decision-making (e.g., by manipulating purchase decisions). Recent advances in Artificial Intelligence (AI) and the explanations of its decisions through explainable AI (XAI) have opened new opportunities for mitigating biased decisions. So far, the potential of these technological advances to overcome anchoring bias remains widely unclear. To this end, we conducted two online experiments with a total of N=390 participants in the context of purchase decisions to examine the impact of AI and XAI-based decision support on anchoring bias. Our results show that AI alone and its combination with XAI help to mitigate the negative effect of anchoring bias. Ultimately, our findings have implications for the design of AI and XAI-based decision support and IS to overcome cognitive biases.

Motivation & Objective

• To examine whether AI and XAI-based decision support can reduce the impact of anchoring bias in consumer decision-making.
• To investigate the individual and combined effects of AI and XAI on mitigating cognitive biases in purchase decisions.
• To provide design implications for ethical information systems that counteract manipulative bias exploitation.

Proposed method

• Conducted two online experiments with a total of N=390 participants to simulate real-world purchase decision contexts.
• Used controlled scenarios where participants were exposed to anchoring cues (e.g., high or low initial prices) to induce bias.
• Implemented AI-based decision support that provided recommendations based on objective data, reducing reliance on initial anchors.
• Integrated XAI components to explain AI recommendations, enhancing transparency and user trust.
• Measured decision outcomes and bias levels through post-decision surveys and behavioral metrics.
• Applied statistical analysis to compare decision quality and anchoring effects across conditions (control, AI-only, AI+XAI).

Experimental results

Research questions

• RQ1 Does AI-based decision support reduce the influence of anchoring bias in purchase decisions compared to no support?
• RQ2 Does the addition of XAI explanations further reduce anchoring bias beyond AI alone?
• RQ3 How do users perceive and respond to AI recommendations when they are transparently explained via XAI?

Key findings

• AI-based decision support significantly reduced anchoring bias compared to the control condition, where participants were not supported.
• The combination of AI and XAI led to a further reduction in anchoring bias, outperforming AI alone.
• Participants exposed to XAI explanations showed higher decision accuracy and lower susceptibility to initial anchors.
• The presence of explanations increased user trust in AI recommendations, contributing to more rational decision-making.
• Both AI and XAI were effective in reducing reliance on heuristic cues, even when strong anchoring cues were present.
• The results suggest that XAI enhances the cognitive de-biasing effect of AI by improving user understanding and reducing over-reliance on initial information.