BiasEye: A Bias-Aware Real-time Interactive Material Screening System for Impartial Candidate Assessment

TL;DR

BiasEye tackles cognitive biases in material screening by integrating a bias-aware visualization system with a four-step pipeline ($PREVENTING\rightarrow\DISCOVERING\rightarrow\LOCATING\rightarrow\MITIGATING$). The approach leverages ChatGPT-4 for information extraction and Ranking SVM-driven insights to model individual reviewer preferences, enabling real-time bias awareness and decision support. In a mixed-design study with 20 participants, BiasEye increased bias awareness, reduced inconsistencies in Phase II, and improved confidence in screening outcomes, albeit with higher cognitive load from added features. The work demonstrates the value of human-in-the-loop, visualization-aided decision-making for impartial screening and outlines design considerations for extending bias mitigation to broader screening contexts.

Abstract

In the process of evaluating competencies for job or student recruitment through material screening, decision-makers can be influenced by inherent cognitive biases, such as the screening order or anchoring information, leading to inconsistent outcomes. To tackle this challenge, we conducted interviews with seven experts to understand their challenges and needs for support in the screening process. Building on their insights, we introduce BiasEye, a bias-aware real-time interactive material screening visualization system. BiasEye enhances awareness of cognitive biases by improving information accessibility and transparency. It also aids users in identifying and mitigating biases through a machine learning (ML) approach that models individual screening preferences. Findings from a mixed-design user study with 20 participants demonstrate that, compared to a baseline system lacking our bias-aware features, BiasEye increases participants' bias awareness and boosts their confidence in making final decisions. At last, we discuss the potential of ML and visualization in mitigating biases during human decision-making tasks.

Figures (13)

• Figure 1: An illustration of the contrast bias emerges in sequential material screening tasks. In such scenarios, condition of adjacent application materials can influence a reviewer's assessment, resulting in reviewers making inconsistent judgments about the same application X under varying conditions.
• Figure 2: The front-end design of BiasEye.
• Figure 3: The overview of data processing and backend model pipeline of BiasEye.
• Figure 4: A visualized indicator of the Statistical view.
• Figure 5: A stacked time bar Ex-situ Table.