Culture & Research Philosophy
AIT Brain Lab studies humans, artificial intelligence, and technologies that improve human lives. We combine experiments, computational methods, and theory to understand problems and develop useful solutions. We value curiosity, careful reasoning, reproducible work, and honest reporting of what works and what does not.
Bring a question you care about; we develop the research direction together. We help assess prior work and set a scope that fits your preparation, interests, degree, and time. Guidance is structured early; you take greater ownership as your skills develop.
PhD and Master’s Expectations
PhD: Develop an Original Contribution
PhD students develop an original contribution to knowledge and become increasingly independent researchers. A contribution may be a new explanation, method, interaction technique, benchmark, dataset, or system that produces meaningful new knowledge through rigorous evaluation.
- Develop a focused research question within the lab’s areas of expertise, together with the supervisor.
- Establish what prior work leaves unanswered and why answering it matters.
- Design and carry out studies that support a clear contribution, including careful analysis of limitations.
- Gradually lead research decisions, writing, and communication of the work.
Originality can be a focused, valuable advance. It does not require a breakthrough at the start, and the supervisor may provide the initial direction.
Master’s: Own a Focused Project within a Larger Effort
Master’s students will usually join an ongoing PhD project or a grant-funded research project. This provides a clearer starting point, shared resources, and support from the research team.
- Agree on a bounded piece of work with clear learning goals and deliverables.
- Contribute through implementation, experiments, data preparation, evaluation, or a focused research question.
- Understand and justify the methods used, evaluate the results, and explain the project’s limitations.
- Take ownership of an identifiable contribution suitable for the student’s degree requirements.
Students contribute as developing researchers with their own learning goals and credit for their work. A related topic proposed by the student is also welcome when supervision and resources are available.
Shared standards, different scope. Both pathways require integrity, thoughtful evaluation, and clear communication. The scale of the project and the expected level of independence differ. Project roles, individual contributions, and authorship expectations should be discussed early and revisited as work develops. These lab expectations complement the formal requirements of each degree programme.
Research Areas
PhD projects develop original contributions within the following four areas. Master’s projects usually address a focused part of these broader efforts. Projects may connect more than one area.
Human–AI Interaction
How can we understand people through their interactions with AI, and design tools that strengthen human agency, judgment, and capabilities?
- Simulated humans: Model personalities, cognitive biases, memory, cognitive load, and other human characteristics; examine when simulations reflect real human behaviour and where they fail.
- Human biases and AI behaviour: Study how people and AI influence one another, including sycophancy, trust, negotiation, and decision-making.
- Tools for human agency and well-being: Develop and evaluate interactions that support human control, cognition, attention regulation, habit formation, and independent judgment.
Related work: Human–AI negotiation · Personality and sycophancy · Attention regulation
Reliable and Adaptive Language Models
How can language-model systems produce more verifiable, constrained, and context-appropriate responses, especially in settings where errors have serious consequences?
- Ontologies and symbolic constraints: Combine LLMs with structured knowledge, rules, and validation to make outputs more predictable and checkable. Evaluate the remaining failure modes.
- Benchmarks and evaluation: Develop tasks and datasets that reveal meaningful capabilities, reliability limits, and failures that existing evaluations miss.
- Adaptive responses: Investigate when a system should adjust its explanation, ask for clarification, retrieve evidence, or abstain, based on the task, user needs, and uncertainty.
Related work: PolicyKG: institutional policies and SHACL knowledge graphs
Medical Vision–Language Learning
Our current deep-learning interest is medical visual question answering (VQA): answering questions about medical images using both visual and language information.
- Methods: Improve how models connect image evidence with questions and answers.
- Data and evaluation: Develop clinically grounded datasets and assess accuracy, robustness, generalisation, and clinically meaningful errors.
- Reliable assistance: Study evidence grounding and uncertainty, including when a model should defer to a clinician.
Physiological Computing and Biomedical Sensing
How can signals from the brain and body support useful interfaces and health-related measurement? This area brings together signal processing, machine learning, sensing, and interaction design.
- EEG and brain–computer interfaces: Develop and evaluate brain-driven communication tools, including SSVEP, P300, and hybrid BCI spellers.
- Raman spectroscopy: Investigate sensing and modelling approaches for non-invasive glucose measurement, with attention to calibration, confounds, and validation.
Demo: BCI speller
Choosing a Project and Sharing the Work
Students from different disciplines are welcome. Together, we agree on the question, responsibilities, support, and milestones around your interests and goals.
Applications and prototypes can be valuable research tools and project outcomes. For a research contribution, we also ask what new capability, evidence, or understanding the work provides beyond building the application.
We aim to share strong research in suitable journals and conferences, including venues such as CHI, ACL, NeurIPS, and relevant specialist venues. Publication plans depend on the contribution, the student’s goals, and programme requirements. We expect careful work and thoughtful revision; acceptance decisions are outside our control.
See our Research Guide for practical help with developing questions, evaluating novelty, planning studies, and communicating results.