⚠️ Under development: inaccurate information shown.

Uncertainty, Explainability, Transparency and Bias in AI

Cohort 1

Cohort 2

Cohort 3

Dealing with interpretive complexity.

Artificial intelligence increasingly shapes critical decisions in healthcare, employment, criminal justice, and civic life – yet these systems often operate as opaque black boxes, perpetuating biases and marginalising vulnerable communities. When a loan application is denied, when a CV is filtered by an algorithm, when medical diagnoses are automated, citizens deserve to understand how these decisions are made and challenge them when necessary. How can we build AI that genuinely serves diverse publics rather than reinforcing existing inequalities?

This research theme explores uncertainty, explainability, transparency, and bias through the lens of interpretive depth and human agency. Drawing on the “Doing AI Differently” initiative (Alan Turing Institute, 2025), we position humanities perspectives and participatory methods as essential, not supplemental, to technical innovation. We investigate how AI can represent multiple valid perspectives rather than imposing monolithic worldviews, how systems can communicate their limitations honestly to calibrate appropriate trust, and how affected communities can exercise genuine decision-making power over technologies that impact them.

Our research addresses urgent challenges at the intersection of HCI, policy, and practice. Current generative AI systems homogenise non-Western writing towards Western norms and privilege North American cultural contexts. Can we design alternative architectures that preserve cultural plurality? Technical transparency doesn’t guarantee human understanding. What makes an explanation meaningful to diverse stakeholders with varying levels of expertise and cultural backgrounds? Bias mitigation techniques often fail to address structural inequities. When should we refuse to deploy AI rather than attempting technical fixes for fundamentally unjust applications?

We aim to develop interpretive technologies that engage with cultural complexity, participatory governance frameworks where communities shape AI systems rather than merely being consulted, and evaluation methods that assess both cultural sensitivity and technical performance. This work is closely aligned with the HCI research communities (ACM CHI, IUI, FAccT), the UK’s principles-based AI governance framework, and the EU AI Act’s requirements for transparency and human oversight.

Place-based and Regional Context

The North East provides a distinctive context for citizen-centred AI research, combining industrial heritage with contemporary economic challenges and strong community organising traditions. Significant socioeconomic inequality, ageing populations in coastal areas, diverse ethnic communities in urban centres, and rural-urban divides create contexts where AI impacts are unevenly distributed. Research here can surface perspectives often marginalised in technology design—healthcare patients navigating algorithmic triage, workers facing automated management systems, residents affected by “smart city” deployments, and cultural heritage organisations preserving diverse narratives.

Advanced manufacturing, healthcare systems, creative industries, and public sector organisations face AI adoption decisions with significant equity implications. Regional SMEs lack resources for implementing responsible AI, creating opportunities for participatory design research that centres on affected communities from the outset. The North East’s vibrant digital inclusion networks, community arts organisations, patient advocacy groups, and cooperative development infrastructure provide research partners interested in co-designing accountable AI systems that genuinely serve local needs.

Relevant Partner Organisations

Our regional ecosystem brings together diverse organisations committed to responsible technology development. Digital innovation organisations in the cultural sector explore how AI tools can enhance rather than undermine creative expression and cultural diversity. Voluntary sector networks connect thousands of community organisations, providing channels for participatory research with groups affected by AI in public services. Health and wellbeing initiatives work at the intersection of arts, health, and technology, investigating how AI can support patient agency whilst addressing uncertainty communication in contexts of incomplete medical knowledge.

The region’s sports development sector engages with questions of algorithmic fairness in performance analytics and talent identification, where bias perpetuates structural inequalities. Genomics and life sciences centres explore responsible health AI, whilst cultural heritage organisations (including archives, museums, and community trusts) examine how AI can preserve diverse narratives without flattening cultural complexity. Advanced manufacturing networks and housing associations investigate transparency requirements in operational AI systems, while corporate partners provide industry perspectives on implementing trustworthy AI, navigating intellectual property and competitive pressures.

This ecosystem enables research that bridges academic rigour with real-world impact, ensuring our work addresses genuine challenges faced by communities, organisations, and citizens navigating an increasingly automated world.

Related Articles & Reading