Taming AI - Matt Jones

19 snips

May 5, 2026

Matt Jones, a Swansea University computer scientist focused on human-centred and inclusive AI, explores ways to tame AI through design and governance. He contrasts fear-based control with understanding, reframes AI as a power like fire, and examines legibility, explainability, data bias, guardrails, human-in-the-loop limits, and the need for inclusive, participatory design.

Ask episode

AI Snips

Chapters

Books

Transcript

Episode notes

ADVICE

Prioritise Legibility Over Only Law

Make AI legible so users can read what a system is doing, because laws can demand behaviour but not show how a system operates.
Matt Jones emphasises users need transparency to assess trust and control, not just regulation.

ANECDOTE

Training Data Reveals Cultural Biases

Jones runs a prompt experiment showing AI reflects biased training data: 'doctor' often evokes a man and generated images of Dharavi were stereotyped by slum-tourist photos.
These examples show models amplify historical and photographic biases from their training sets.

ADVICE

Require Explainability For Model Claims

Demand explainability: use techniques like SHAP and LIME to surface which features drive AI decisions so you can judge credibility.
Jones shows an animal classifier that uses snow as the salient feature, misleading its wolf/dog labels.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

Watch the Q&A session: https://youtu.be/gj4d75_Clgg

In this lecture, we look at proposals to limit AI powers and impacts, so bad outcomes are outweighed by social benefits from the technology. I’ll explain design processes (such as Human-Centred AI and Responsible AI) and technological approaches for AI system qualities like trustworthiness, explainability and “human in the loop”. We will explore how we, as individuals, can use AI based systems in discerning ways; and look at what governments can do to help their citizens thrive in an AI-future.

This lecture was recorded by Professor Matt Jones on the 21st of April 2026 at Barnard’s Inna Hall, London

Matt Jones is a computer scientist at Swansea University - and a Fellow of the British Computer Society - who works alongside colleagues from many other disciplines and directly with everyday folk across the world to explore the future of digital technologies. Over the last 30-plus years, this human-centred approach has led to novel approaches for, amongst other things, mobile phone-based information searching and browsing, pedestrian navigation, voice assistants and deformable displays.

Much of his work has been driven by intense and sustained engagements with “low resource” communities from informal settlements in India, South Africa, and Kenya. Through their generous and gracious participation, these extra-ordinary users with the fresh and diverse perspectives have stimulated insights into the future of digital technologies for everyone, globally. In all this work, Matt works as part of a long-standing collaborative team with Jen Pearson, Simon Robinson and Thomas Reitmaier (from Swansea) and colleagues in India (including Dani Raju) and South Africa (including Minah Radebe).

His work has been supported by the UK’s science funders (EPSRC and UKRI). Currently, this funding includes a Fellowship to explore the future of interactive AI and leadership roles in responsible AI and inclusive digital technologies. This funding has led to a series of impactful publications, talks and influences on people, policies, and practices.

Matt has collaborated with private, public and third sector organisations, including Microsoft, the NHS, Google, IIT-B, the BBC and IBM. He is a member of the Foreign and Commonwealth Development Office’s Research Advisory Group and Welsh Government’s AI reviews.

The transcript and downloadable versions of the lecture are available from the Gresham College website: https://www.gresham.ac.uk/watch-now/ai-taming

Gresham College has offered free public lectures for over 400 years, thanks to the generosity of our supporters. There are currently over 2,500 lectures free to access. We believe that everyone should have the opportunity to learn from some of the greatest minds. To support Gresham's mission, please consider making a donation: https://gresham.ac.uk/support/

Website: https://gresham.ac.uk
Twitter: https://twitter.com/greshamcollege
Facebook: https://facebook.com/greshamcollege
Instagram: https://instagram.com/greshamcollege

Support the show