The mission would be to build consensus and energy around the opportunity for an EU-wide activity on AI for science – AI-enabled accelerated scientific discovery across the sciences. This would build on Europe’s heritage in the scientific method, leading basic science and AI research institutes, as well as a well-established lab instrument sector as an export hidden champion.
The TG would aim to be open the concerns of all parts of the ADRA communities:
• AI – foundational models, simulation tools and digital twins, knowledge extraction from the scientific literature, multi-agent systems, human-centric AI & UI
• Big data – including multi-modal, spatio-temporal, natural language, representation
• Robotics – for new concepts in lab automation, field robotics, manipulation, digitalisation
AI Scientists (also known as 'Self-driving Labs') are AI systems integrated with laboratory robotics to enable cycles of closed-loop automation of scientific research. AI is used be used to form hypotheses, design experiments, and intelligently control laboratory robotics. Foundation Models are set to revolutionise scientific research by learning from vast datasets and applying the versatile and easily adaptable to new tasks and problems, even without further training. Large Language Models (LLMS) can be used to extract explicit knowledge from the scientific literature. Once the knowledge is it is easier to use by human and AI scientists than the original text. Ultimately, all the knowledge in scientific papers can be extracted and then there is no need for the text (and copyright).
The TG would work to identify the most promising use-cases, around which we might consider ‘challenges’ (the vaccine machine, the smart lab assistant) And describe ways in which fundamental research might be accelerated with the use of AI and laboratory robotics. The scope could include observational and experimental sciences as well as system sciences and be applicable to part of the humanities.
