Trustworthy AI systems require more than performance alone. This research theme examines how AI can be made explainable, transparent, safe, robust, fair, and accountable in settings where decisions affect institutions, markets, and public life. We ask how technical design, evaluation, and governance can reduce risk, improve oversight, and support responsible adoption. The work brings together scientific rigor and societal relevance, helping shape AI systems that are not only powerful, but also understandable, dependable, and aligned with public values.
Trustworthy AI Systems
Advancing explainable, safe, fair, and accountable AI through rigorous research, evidence, and collaboration.
Explore researchResearching AI people can trust
Core pillars of the theme
Explainability and transparency
We study how AI systems can be inspected, interpreted, and communicated clearly to users, policymakers, and experts. Transparency supports better oversight, more informed use, and stronger public confidence.
AI safety and robustness
Our research examines how AI behaves under uncertainty, stress, and shifting conditions. The goal is to improve resilience, limit failure modes, and strengthen reliability in real-world deployment.
Fairness
We investigate how bias can emerge in data, models, and outcomes. This work supports more equitable AI by identifying risks of unequal treatment and by informing better design and evaluation practices.
Accountability and governance
We explore the structures needed to assign responsibility, enable oversight, and guide responsible use. Governance-focused research helps align AI systems with institutional standards, legal expectations, and ethical principles.
Impact in focus
4Core research pillars shaping the theme
AIPrimary focus area for trustworthy system design
EvidenceResearch outputs informing practice and policy
Current research directions
Current work spans methods for interpreting model decisions, approaches for stress-testing AI performance, and frameworks for evaluating fairness and accountability across contexts. Research also addresses governance questions around oversight, auditability, and responsible deployment. Together, these directions help translate trustworthy AI principles into practical guidance for researchers, organizations, and decision-makers.