Research
Research Program
My research program integrates theoretical frameworks, evaluation methodologies, and empirical studies to investigate how humans and AI interact, collaborate, learn, reason, and make decisions under uncertainty. It connects Interaction Quality, Meaningful Human Oversight, and Responsible AI Governance across education and healthcare, with conversational and socio-technical AI as cross-cutting contexts. The overarching aim is to inform the design and governance of AI systems that strengthen rather than substitute human capabilities.
Collaboration
Quality
Oversight
Governance
Research architecture
Three pillars, connected by one goal.
The program links Human–AI evaluation, meaningful oversight, and responsible governance across education, healthcare, and conversational or socio-technical settings.
Human-Centered AI for effective Human–AI collaboration
Preserving agency · expertise · critical thinking · responsibility
Interaction Quality & Human–AI Evaluation
Multidimensional evaluation · conversational agents and LLMs · AI-supported decision-making.
Meaningful Human Oversight
Calibrated reliance · automation bias · cognitive offloading · deskilling · expertise preservation.
Responsible AI Governance & Deployment
Validation · methodological quality · transparency · standardization · accountability · deployment.
The three pillars
Concepts, methods, and applications.
Each pillar addresses a different layer of the same research problem, from the quality of interaction to the preservation of human capabilities and the conditions for trustworthy deployment.
Interaction Quality & Human–AI Evaluation
Developing theoretical, methodological, and psychometric frameworks for evaluating Human–AI interaction beyond system performance alone, integrating cognitive, behavioral, collaborative, and experiential dimensions.
Meaningful Human Oversight
Investigating when and how AI can augment human judgment while preserving agency, competence, expertise, and responsibility, with attention to calibrated reliance, overreliance, automation bias, cognitive offloading, and AI-induced deskilling.
Responsible AI Governance & Deployment
Translating Human-Centered AI principles into rigorous evaluation, validation, transparency, standardization, socio-technical accountability, and responsible real-world deployment.
Application domains
Education and healthcare are the primary application domains; conversational agents, Large Language Models, and AI-supported professional or organizational workflows provide cross-cutting contexts.
Research trajectory
From methodological foundations to an integrated agenda.
The trajectory connects earlier work in predictive modelling, complex systems, explainability, and AI in education to the current Human-Centered AI program and its emerging research directions.
Predictive modelling & complex systems
- Forecasting and data-driven modelling
- Social dynamics and complex systems
- Quantitative and computational methods
Explainable AI, education & collaboration
- Explainable recommender systems
- Personalized learning
- Teacher–AI collaboration
- Explainability reviews & educational AI
Evaluation & high-stakes contexts
- Methodological quality in Medical AI
- Broader AI & society perspective
- Human–AI configurations
- Collective cognition and adjunct AI
Integrated Human-Centered AI
- Interaction Quality
- Meaningful Human Oversight
- Responsible AI Governance
- Education and healthcare applications
Emerging directions
Parallel extensions of the current program
Cognitive Sustainability
Long-term preservation and development of human cognitive capabilities in AI-mediated environments.
Human–AI Co-Agency
Meaningful distributions of initiative, expertise, autonomy, and responsibility.
Longitudinal Human–AI Interaction
Sustained effects on reliance, competence, learning, judgment, and behavior.
Advanced Human–AI Evaluation
Richer methodological and psychometric evaluation of interaction and human outcomes.
Selected outcomes & recognition
Evidence of trajectory
Publication profile, venues, authorship roles, and selected competitive recognition complement the conceptual architecture of the research program.