Research

Human-AI Collaboration in High-Risk Decision Domains with Eye Tracking

Modelling human behaviour in high-risk decision workflows using eye-tracking data — the empirical arm of the PhD, spanning six papers across IJHCI, TOCHI, TE, and INCOSE CAS.

Status: Active
Period: 2023 – Present
Publications: 6

Yichuan wearing eye-tracking glasses, interacting with a Human-AI electricity-demand forecasting interface (SHAP interpretability + adjustment panel)

Outputs — 6 papers

Calibrated Intervention in Human-AI Collaborative Forecasting: How Modification Intensity and Direction Interact to Determine Performance
Published · International Journal of Human-Computer Interaction
Process-Informed Crowd Selection: Recovering Collective Prediction Gains Under AI Anchoring Through Eye-Tracking-Based Evidence Engagement Estimation
Preprint · Under Review · ACM Transactions on Computer-Human Interaction (TOCHI)
Behavioral Adaptation in Human-AI Decision Making: A Longitudinal Study Using Synchronized Eye and Mouse Tracking
Accepted · 33rd Int. Conf. Transdisciplinary Engineering (TE2026)
Bridging Human Cognition and AI Systems: A Transdisciplinary Engineering Study of Temporal Interaction Patterns in Collaborative Forecasting
Accepted · 32nd Int. Conf. Transdisciplinary Engineering (TE2025)
Multimodal Analysis of Human–AI Collaboration in Adaptive Forecasting in Electricity Demand Prediction
Accepted · INCOSE Complex Adaptive Systems
Human Decision Making Assisted by Artificial Intelligence: Electricity Demand Forecasting in Japan
Published · Advances in Transdisciplinary Engineering

Challenge

AI-only forecasting systems struggle under sparse data, unexpected events, and shifting regimes. Expert forecasters routinely adjust model outputs using domain knowledge, yet current systems rarely support structured collaboration between humans and AI.

Approach

I design interactive decision-support interfaces implementing an AI-first paradigm, where ML models produce initial forecasts that experts refine. Cognitive state transitions and interaction strategies are modelled with sequence models over multimodal features (gaze, mouse trajectories, screen interaction).

Results

Human–AI collaboration improved forecast accuracy under sparse and volatile conditions compared to AI-only baselines. Findings consolidated into a human-centered collaboration framework; results disseminated via IJHCS submission and TE2025/INCOSE(CAS) acceptances.