Fourth Workshop on Multimodal AI

11–12 June 2026, UCL East Campus, One Pool Street, London, UK

Latest news:

Access to the hackathon webinar recording and slides.
Accepted abstracts posted
Tentative programme now available
Sponsorship opportunities
Early bird registrations now closed
Call for abstracts now closed
Travel bursaries now closed
Volunteer opportunities now closed

We are pleased to invite you to join us for our Fourth Workshop on Multimodal AI (MultimodalAI’26), taking place on 11–12 June 2026 at the UCL East Campus, One Pool Street, London. This workshop was an in-person-only event, preceded by a full-day hackathon on 10th June.

Multimodal AI integrates diverse types of data, including text, images, audio, time series, graphs, and more. It is rapidly transforming how we engage with information and technology. MultimodalAI’26 will bring together researchers and practitioners from AI, data science, and related scientific and applied fields to discuss challenges, share innovative solutions, explore future collaborations, and strengthen the UK’s growing multimodal AI community.

The workshop is hosted by the UK Open Multimodal AI Network (UKOMAIN), a £1.8 million EPSRC Network Plus initiative. The programme will feature:

Keynote presentations by leading experts
Community pitches, talks, and poster sessions
Engaging panel discussions
Updates from open multimodal AI benchmark projects
Funding and policy briefings
A full-day pre-workshop hackathon on 10 June
Travel bursaries and volunteering opportunities to support inclusive participation

Prizes will be awarded for outstanding talks, pitches, posters, and hackathon contributions.

Keynote Speakers

Aron Walsh

CSO at CuspAI and Professor in Materials Design at Imperial College London

Chris Barnes

Head of Science of AI at National Physical Laboratory and Professor of Systems and Synthetic Biology at UCL

Henry Lee

CEO at Cultivarium

Julia Hirschberg

Professor of Computer Science at Columbia University

Tom Pollard

Research Scientist at Massachusetts Institute of Technology and Technical Director of PhysioNet

Zoe Kourtzi

Professor of Cognitive Computational Neuroscience at University of Cambridge and CSO at Prodromic

Time	Event
09:30 – 18:00	MultimodalAI'26 hackathon
10:00 – 17:00	MultimodalAI'26 workshop early registration

Final Programme

Workshop Day 1: 11 June 2026

Time	Event
09:30 – 10:00	Registration, coffee/tea, and biscuits
10:00 – 10:15	Welcome address: Amanda Brock, CEO at OpenUK
10:15 – 10:30	Opening talk: Haiping Lu, Director, UK Open Multimodal AI Network
10:30 – 12:10	Session 1 - Chairperson: Nicola Morley, University of Sheffield
10:30 – 11:10	Keynote 1: Aron Walsh, CSO at CuspAI and Professor in Materials Design at Imperial College London
	Title: Materials on Demand
	Abstract: The inverse design problem (given a target property or function, identify the optimal material) represents one of the central challenges in materials science. The landscape of materials theory and simulation is addressing this problem through the integration of new techniques and tools from the artificial intelligence (AI) community. Progress in hardware, including classical supercomputers and emerging quantum computers, alongside software advancements incorporating advanced algorithms and statistical machine learning models, is expanding what is now possible. A particular opportunity lies in multimodal AI, which can bridge heterogeneous data streams spanning computation, synthesis, and characterisation to build richer and more transferable representations of materials. Recent developments, such as large language models and generative diffusion techniques, are unlocking application areas ranging from multimodal characterisation to integration with self-driving laboratories. I will survey the evolution of data-driven approaches to materials on demand, highlighting their potential to expedite the identification of compounds essential for the next generation of clean energy technologies. The talk will close with reflections on the translation of academic research into emerging industry, with particular attention to the growing AI-for-materials ecosystem.
11:10 – 11:45	Community talks 1
11:45 – 12:10	Group photos
12:10 – 13:20	Lunch and posters
13:20 – 14:50	Session 2 - Chairperson: Dezong Zhao, University of Glasgow
13:20 – 14:00	Keynote 2: Julia Hirschberg, Professor of Computer Science at Columbia University
	Title: Radicalization and De-Radicalization in Social Media
	Abstract: Our prior work on radicalization focuses on Increasing our understanding of online radicalization efforts from analyzing right- and left-leaning group videos in social media including groups from outside the USA. We studied what viewers found the videos appealing, what techniques were employed in producing radical content, and how we could computationally identify these techniques. More recently we have worked to identify potentially effective de-radicalizing techniques and worked to build a unified multimodal processing pipeline toolkit to help us create conversational systems on social media that are useful in de-radicalizing viewers.
14:00 – 14:30	Community talks 2
14:30 – 14:50	Hackathon presentation
14:50 – 15:50	Coffee/tea, biscuits, and posters
15:50 – 17:30	Session 3 - Chairperson: Richard Gilham, Bristol Centre for Supercomputing
15:50 – 16:30	Keynote 3: Chris Barnes, Head of Science of AI at National Physical Laboratory and Professor of Systems and Synthetic Biology at UCL
	Title: An Overview of AI at NPL
	Abstract: The National Physical Laboratory (NPL) is the UK’s national metrology institute. It develops and maintains the country’s primary measurement standards and provides the measurement science, calibration, testing, and scientific expertise that underpin industry, innovation, and public services—helping ensure measurements are accurate, comparable, and trustworthy. In this talk, Chris Barnes (Head of Science for AI, NPL) will give an overview of NPL’s AI activity and how it supports the UK’s ambitions for trustworthy and deployable AI across industry. The talk is structured into three parts covering the main aspects of AI@NPL. First, Trustworthy AI: how measurement thinking translates into AI practice, covering concepts such as uncertainty, robustness, interpretability, and data quality, and how these properties can be characterised, tested, and compared in a repeatable way. Second, AI for Metrology: examples of how modern machine learning can be used to enhance the scientific process. Third, AI assurance: the emerging methods, standards, and evaluation approaches needed to build justified confidence in AI systems for critical applications.
16:30 – 17:30	Research Panel: From Models to Evidence: Making Multimodal AI Deployable
	Question 1: What community outputs would most help multimodal AI move from research promise to real-world deployment?
	Question 2: What evidence should researchers provide before claiming a multimodal AI system is trustworthy enough for real-world use?
17:20 – 17:30	Hackathon winner announcement
17:30 – 20:00	Networking Reception: UKOMAIN Connections Corners
	Informal networking around three themes: • Funding, Proposals & Collaborators • Evidence, Evaluation & Responsible Deployment • Careers, Mentoring & Inclusive Leadership

Workshop Day 2: 12 June 2026

Time	Event
08:30 – 09:00	Registration, coffee/tea, and biscuits
09:00 – 10:25	Session 4 - Chairperson: Nataliya Tkachenko, Lloyds Banking Group
09:00 – 09:20	Invited talk: Andy Lawrence, Head of Engineering at EPSRC
09:20 – 10:00	Keynote 4: Henry Lee, CEO at Cultivarium
	Title: Scaling Biological Expertise
	Abstract: Cultivarium builds scientific tools that turn biological discovery into real-world capability. Remarkable genetic capabilities and chemistries useful for beneficial biotechnologies can be found in natural biological systems, yet nearly all of them sit outside today’s laboratory. To make the vast number of non-model organisms available for reproducible laboratory study, we have built a technology platform for domesticating them, compressing the typical time frame from decades to days. In this talk I will trace our journey from deterministic software to AI tooling for accelerating scientific discovery, biological engineering, and the real-world deployment of knowledge systems. Hermes, our large language model (LLM) harness, helps scientists navigate existing knowledge, generate hypotheses, and run analyses across iterative discovery campaigns. PRISM, our vision-language model (VLM) harness, captures experimental work as audio and video feeds to generate traceable documentation and surface tacit knowledge that is otherwise earned one scientist at a time. These multimodal AI tools form the basis for a new software stack for embedding and distributing expertise, accelerating progress in frontier biology research programs.
10:00 – 10:25	OMAIB project showcase
10:25 – 11:25	Coffee/tea, biscuits, and posters
11:25 – 13:00	Session 5 - Chairperson: Fei He, Coventry University
11:25 – 12:05	Keynote 5: Zoe Kourtzi, Professor of Cognitive Computational Neuroscience at University of Cambridge and CSO at Prodromic
	Title: Multimodal AI for Early Dementia Prediction: From Cloud to Clinic
	Abstract: Early prediction of brain (e.g. neurodegenerative) disorders is key for clinical management and patient outcomes. Predicting whether individuals with mild cognitive impairment or people without symptoms will decline or remain stable is impeded by patient heterogeneity due to comorbidities, lifestyle and disease severity. Despite the importance of early diagnosis of dementia for prognosis and personalised interventions, we still lack robust tools for predicting individual progression. We propose a novel clinical AI predictive prognostic modelling approach that mines multimodal data to derive an individualised prognostic marker of cognitive decline at early stages of dementia or before symptoms occur. We validate our approach against routinely collected real-world patient data from memory clinics over time, showing that our clinical AI marker is more sensitive than the standard of care (cognitive tests, MRI scans). Our clinical AI approach has strong potential to facilitate effective patient stratification into clinical pathways and trials, reducing patient misdiagnosis and enhancing trial efficiency with important implications for clinical translation and drug discovery.
12:05 – 12:15	Invited talk: Jake Bowden, Bioinformatician at Twig Bio
	Title: Canopy: A Heterograph Foundation Model for Metabolic Engineering
12:15 – 12:25	Invited talk: Farah Shamout, Assistant Professor of Computer Engineering at NYU Abu Dhabi
	Title: Multimodal AI for Patient-Centered Healthcare Technologies
12:25 – 13:00	Community talks 3
13:00 – 14:00	Lunch and posters
14:00 – 16:00	Session 6 - Chairperson: Peter Charlton, UKOMAIN Executive Board
14:00 – 14:40	Keynote 6: Tom Pollard, Research Scientist at Massachusetts Institute of Technology and Technical Director of PhysioNet
	Title: Data for Health AI in a Time of Change
	Abstract: We will explore how the health AI community is moving from an era focused mainly on model building to one that must confront deeper questions about data quality, access, governance, and representation. Using examples from recent controversies in medical AI and from the development of open research resources such as PhysioNet and MIMIC, the talk will argue that progress now depends on creating data ecosystems that are more trustworthy, more reusable, and more faithful to the realities of clinical care.
14:40 – 15:30	Career Panel: Building a Sustainable Research Career in Multimodal AI
	Question 1: How should early-career researchers decide which leadership, service or network opportunities to take on — and which to decline?
	Question 2: Beyond papers, what skills will matter most for the next generation of multimodal AI researchers?
15:30 – 15:50	Presentation winner announcement
15:50 – 16:00	Closing remarks
16:00 – 17:00	Post-workshop networking

Key Dates and Deadlines

		Close Date
Call for Abstracts		Wed, 6th May

Abstract Acceptance Notification		Mon, 11th May

Early Bird Registration		Closes when full or Mon, 18th May

Cancellation Deadline (with refund)		Tue, 26th May

Call for Volunteers		Wed, 6th May

Volunteer Acceptance Notification		Mon, 11th May

Travel Bursary for Inclusive Participation		Wed, 6th May

Travel Bursary Decision Notification		Mon, 11th May

Organising Committee

Haiping Lu

Director of the UK Open Multimodal AI Network, Professor of Machine Learning & Head of AI Research Engineering, University of Sheffield

Nicola Morley

Deputy Director of the UK Open Multimodal AI Network & Professor of Materials Physics

Peter Charlton

Lead for Early Career Researchers at the UK Open Multimodal AI Network & Senior Research Scientist, Nokia Bell Labs

Nataliya Tkachenko

Lead for Industry & Public Engagement at the UK Open Multimodal AI Network & Generative AI Ethics & Assurance Lead, Lloyds Banking Group

Yao Zhang

Lead for Equality, Diversity, and Inclusion at the UK Open Multimodal AI Network & Lecturer in Marine/Maritime Digitalisation and Automation, UCL

Dezong Zhao

Lead for UKRI and Government at the UK Open Multimodal AI Network & Reader in Autonomous Systems, The University of Glasgow

Aayush Chiniwalla

Network Coordinator of the UK Open Multimodal AI Network

Xianyuan Liu

Senior AI Research Engineer & Assistant Head of AI Research Engineering, University of Sheffield

Partners

Our Key Sponsors

Contact Us

Email the organisers: multimodalai26-group@sheffield.ac.uk

UCL East Campus, One Pool Street, London, E20 2AF