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.
Computers and Society,Artificial Intelligence,Human-Computer Interaction,Machine Learning,General Economics
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is: how to use artificial intelligence (AI) and explainable artificial intelligence (XAI) to overcome anchoring bias, thereby improving individuals' performance in decision - making tasks such as market pricing estimation. Specifically, the researchers explored the impact of AI and XAI support on anchoring bias through two online experiments and evaluated whether these technologies can help individuals reduce the bias caused by anchors and improve the accuracy of decision - making.
### Research Background and Problem Description
1. **Anchoring Bias**: Anchoring bias refers to the fact that when an individual makes a decision or estimate, the initially given information (i.e., the anchor) will significantly affect their subsequent judgment, even if this anchor is irrelevant, arbitrary, or uninformative. This phenomenon is particularly common in purchase decisions. For example, e - commerce platforms often use the pre - discount price as an anchor to induce consumers to make impulse purchases.
2. **Existing Challenges**:
- Currently, there are few studies on the role of AI and XAI in overcoming anchoring bias.
- Although AI can reduce anchoring bias by predicting and reframing the decision - making environment, relying solely on the prediction information provided by black - box models may not be sufficient to completely eliminate the bias.
- XAI can transform the data patterns discovered by complex models into a human - readable form, and theoretically can help people better understand and adjust their own decision - making processes, but its actual effect has not been fully verified.
### Core Objectives of the Paper
- **Evaluating the Support Effects of AI and XAI**: Prove through experiments whether AI used alone and in combination with XAI can effectively reduce the impact of anchoring bias, thereby enhancing individuals' performance in market pricing estimation tasks.
- **Exploring Different Types of Anchors**: Study the impact of price anchors and irrelevant anchors (such as high - price tags in advertisements) on decision - making, and test their interactions in the presence of multiple anchors.
### Experimental Design
To answer the above questions, the author designed two online experiments, involving a total of 390 participants. In the experiments, four decision - support conditions were set (no support, AI - only support, XAI - only support, AI + XAI combined support), and the impact of these changes on participants' decisions was observed by changing the type of anchors (no anchor, price anchor, irrelevant anchor, and their combinations) that appeared in the tasks.
### Conclusions
The research shows that AI used alone and in combination with XAI does help to alleviate the impact of anchoring bias. In particular, in the absence of a clearly provided price, XAI can better help participants estimate a reasonable market price. However, using XAI alone cannot effectively reduce the bias from price anchors or irrelevant anchors. Therefore, the combination of AI and XAI may be one of the key directions for developing more effective decision - support systems in the future.
In summary, this paper aims to reveal the potential of AI and XAI in overcoming cognitive biases and provides new perspectives and empirical evidence for designing better information systems.