Today (Monday 23 October 2023), the British Science Association (BSA) and Google DeepMind have published insights from leaders from across research, industry and policy detailing the current landscape and future potential of artificial intelligence (AI) across science.

Read the full report: Untapping the potential of AI in science

The BSA and Google DeepMind partnered on this project as part of the BSA’s thought leadership programme, For Thought.

It is widely accepted that AI has the potential to transform and accelerate discovery and innovation. This is evident through advances in life sciences, energy, and planetary science. The BSA and Google DeepMind hosted a roundtable inviting colleagues from organisations including The Alan Turing Institute, Department for Business and Trade and the European Molecular Biology Laboratory - European Bioinformatics Institute (EMBL-EBI) to share their experiences and knowledge.

Collectively the 17 leaders concluded that there are challenges specific to AI that affect its use across scientific disciplines. The resulting publication reflects the key insights from the discussion, and is divided into three sections:

• Supporting scientific discovery looks at the ‘hotspots’ for AI in science, the need for a strong pipeline of AI skills, and why expanding access and supporting multidisciplinary collaboration is essential;
• Data and AI for science discusses the key challenges involved in using data for AI in science, including transparency, security, and ethics;
• Where next for AI in science? identifies opportunities for developing the landscape for AI in science, and suggests priorities for supporting continued growth.

The group established five priorities for overcoming these AI-specific challenges:

• Improving the supply of AI talent in science, and introducing certifications to recognise AI skills at all stages of the pipeline;
• Supporting multidisciplinary teams;
• Designing policy frameworks for scientific data transparency, ethics, and sharing; 
• Ensuring accessibility for small businesses and research teams;
• Defining ‘good metrics’ and providing benchmarks.

To truly and fairly untap the potential of AI in science, coordinated action from stakeholders across the research and innovation ecosystem is needed to ensure that appropriate regulation and processes are in place. This infrastructure must be developed by and for people from all backgrounds, rather than a small subset of the scientific population.

The BSA and Google DeepMind hosted this roundtable on 12 September 2023 at The Royal Society. Thank you to those who attended and contributed to this rich discussion.

Find out more about our work with Google DeepMind