AI for Science Summit – driving progress as a community

13 February 2024

Alongside the AI for Science community here in Cambridge, colleagues across other institutions in the network of AI programmes funded by Schmidt Futures are also drawing together communities of like-minded scientists. In December 2023, in collaboration with the Schmidt Futures funded programmes at Imperial College London and the University of Oxford, the Accelerate Programme hosted an AI for Science summit bringing together over 100 researchers working in AI for Science across the UK, US and Asia.

Diverse applications with shared challenges

Across the two days, talks and panel discussions explored a range of applications of AI across diverse disciplines from the use of AI in polar research, work with industry on supply chains to applications in chemical discovery. Despite the diversity of topics explored, a number of shared challenges and opportunities emerged including the challenges of working across disciplines, closing the loop from innovation to implementation and working with incomplete data.

As well as shared challenges, insights on best practice also emerged with unworkshops on large language models and software engineering highlighting the importance of collaboration and openness and clarity.

AI for Science pioneers

Keynote speaker, Shakir Mohamed (Research Director, DeepMind and Associate Fellow, Leverhulme Centre for the Future of Intelligence), explored the role of generative AI in scientific discovery and highlighted the role of AI for Science practitioners as pioneers.

Practitioners are both working at the frontier of new horizons of generative AI for Science and defining what these new horizons are. As this new role of pioneers emerges, coming together as a community provides the opportunity to explore how we can shape the field as we work together to accelerate progress.

Taking the next steps to reach the potential of AI for Science

The final unworkshop drew together participants to consider what steps we can take as a community to achieve the full potential of AI for Science. As well as increased communication within the community, the need to collaborate with domain experts was also highlighted as a crucial step. Building trust is also key, both by improving explainability and by sharing negative results.

Across the two days, both the potential of AI to enable new discoveries and the power of the community to realise this potential were evident. By working together, sharing research and best practice, the AI for Science community can have real impact in enabling the next wave of AI driven discoveries.

Thank you to all of our speakers, panellists and participants for their contributions across the two days of discussion.

If you are interested in taking part in future community events, please contact the team on

Artist Dan Andrews recorded discussions from the day in a series of graphic captures which can be viewed here.

A selection of photographs from the event can be viewed here.