AI safety & trustworthy AI
Evaluate reliability, robustness, and human oversight so that AI systems can be understood, tested, and used responsibly.
Evaluation and calibration · Interpretability · Safety assurance
RESEARCH AT FAIRES
Building knowledge.
Making systems more trustworthy.
Our research agenda connects artificial intelligence, computing, and the methods needed to make complex systems dependable.
FAIRES develops its own research programs, assembles teams, and pursues collaborative and grant-supported work.
We aim to produce peer-reviewed research, reusable software, carefully documented datasets, and educational resources. Our approach emphasizes clear research questions, reproducible methods, and honest reporting of limitations.
We are developing our program portfolio. The themes below define our scientific scope and the kinds of projects we welcome; they do not represent a list of funded awards.
AREAS OF INQUIRY
Research across connected fields.
Multi-agent systems
Study how interacting AI systems coordinate, disagree, and propagate errors—and how oversight can improve their behavior.
Agent interactions · Scalable oversight · System-level evaluation
Cybersecurity & privacy
Investigate threats to data, software, and networks through digital forensics, security analysis, and privacy-aware computing.
Digital forensics · Threat modeling · Network security
Software engineering & assurance
Develop methods for specifying, building, and assessing dependable software, including systems that incorporate AI.
Security requirements · Software assurance · Empirical methods
Computer science & applied computing
Connect foundational methods with real-world questions through machine learning, data science, and computational experimentation.
Machine learning · Data systems · Interdisciplinary applications
Research & computing education
Translate scientific knowledge into educational resources and investigate how people learn, evaluate, and work with computing systems.
Human–AI collaboration · Computing education · Research training
PROGRAM DEVELOPMENT
Directions for the next body of work.
We welcome collaborators and funders interested in shaping focused programs around these questions.
Safety in interacting AI systems
How do errors, incentives, and information boundaries affect groups of AI agents? Potential studies examine coordination, disagreement, failure propagation, and mechanisms for independent review and human oversight.
Evidence for trustworthy human–AI collaboration
When should people rely on an AI judgment, and when should they intervene? Potential work includes reliability and calibration, transparent evaluation workflows, and human-supervised applications in education and other consequential settings.
Assurance for AI-enabled software
How can security and dependability be designed into intelligent systems? Potential work connects security requirements, threat modeling, empirical software engineering, and quantitative evaluation.
A FOUNDATION TO BUILD ON
Research experience.
Documented contributions.
Our founders’ scholarship spans security requirements engineering, AI evaluation, digital forensics, and applied machine learning. Explore selected contributions that inform FAIRES’ scientific direction.
View selected founder scholarshipLET’S ADVANCE THE WORK
Help shape the next research program.
Tell us about your question, methods, team, and opportunities for collaboration.