Explainable legal AI
Interfaces and representations that expose recommendation rationale, source evidence, uncertainty and the boundaries of automated assistance.
PhD research · UKRI CDT-AIMLAC
Making AI-supported legal systems understandable, navigable and accountable.
My research investigates how visual representation, structured knowledge and human-centred AI can help non-experts and professionals reason about legal processes without hiding uncertainty or oversimplifying the law.
Research pillars
A cross-disciplinary programme spanning AI, visualisation and justice.
Legal pathway visualisation
Interactive maps, flow structures and narrative techniques for helping people understand possible routes, decisions, deadlines and institutions.
Grounded retrieval & RAG
Human-curated legal data, semantic retrieval, classification and structured generation for more contextually relevant AI interactions.
Human-centred evaluation
Co-design, expert review, usability testing and evidence-based iteration with legal professionals, advice workers and potential users.
Interoperable digital justice
Data schemas, referral protocols, platform connections and “no wrong door” models for a joined-up ecosystem of legal support.
Responsible innovation
Transparency, inclusion, accessibility, fairness, human oversight and the distinction between information and regulated legal advice.
Method
From domain research to tested technical artefact.
I combine software engineering with design research and empirical evaluation rather than treating legal AI as a model-selection problem alone.
UnderstandLiterature, legal context, existing services and user needs
StructureKnowledge models, categories, schemas and process templates
PrototypeWireframes, storyboards, interaction flows and working systems
EvaluateExpert workshops, user testing, accuracy and usability review
IterateRefine language, pathways, retrieval, explanations and safeguards
Visual forms as part of reasoning—not decoration.
My visualisation work began with narrative and immersive data experiences and developed into research on legal interpretation. Across these domains, the central concern is the same: how can an interface reveal relationships, pathways and uncertainty without misleading the viewer?
Narrative migration visualisation
Interactive maps and visual stories explored refugee movements in the United States, combining data preprocessing, geospatial representation and narrative structure. This work resulted in publications through Eurographics/CGVC.
Immersive WebXR analytics
A web-based immersive visualisation prototype investigated how spatial interaction and multiple views can support data exploration across ordinary browsers and XR devices.
Legal interpretation
Current work studies process maps, relationship diagrams, timelines, visual narratives, knowledge structures and AI-assisted explanations across legal domains. The aim is not simply to make law “look simpler,” but to support careful interpretation while retaining context and user agency.
Policy & public impact
Research connected to live digital-justice questions.
My work contributes to policy-facing discussions on inclusion, explainability, interoperability, consumer protection and responsible AI.
Online Procedure Rule Committee
Digital justice inclusion and pre-action models
Co-authored Bangor University’s consultation response, recommending transparency, human-centred co-design, pathway visualisation, grounded AI, data standards and “no wrong door” interoperability.
Law Society
21st Century Justice
EmployODR was featured as an example of an AI-powered employment-law tool with encouraging expert testing and potential to support access to justice at relatively low operating cost.
Research collaboration
Discuss collaboration Working on legal AI, visual analytics or digital public services?
I welcome conversations with research groups, public bodies and product teams tackling explainability, retrieval, process guidance and responsible AI.