Key Takeaways:
- ARIA has funded 12 projects to develop AI scientists that can conduct novel research and experiments
- The projects include developing AI-powered robots and systems for designing and running experiments in fields such as nanotechnology, chemistry, and materials science
- Each project will receive around £500,000 to cover 9 months’ work and will aim to demonstrate the ability of AI scientists to come up with novel findings
- The funding is part of an experiment by ARIA to determine the current state of the field and inform future funding decisions
- The projects involve teams from universities and industry in the UK, US, and Europe, and will help to advance the development of AI-powered science and research
Introduction to ARIA’s Funding Initiative
There are better uses for a PhD student than waiting around in a lab until 3am to make sure an experiment is run to the end, says Ant Rowstron, ARIA’s chief technology officer. This statement highlights the need for automation and innovation in scientific research, which is exactly what ARIA’s latest funding initiative aims to achieve. ARIA picked 12 projects to fund from the 245 proposals, doubling the amount of funding it had intended to allocate because of the large number and high quality of submissions. This demonstrates the high level of interest and potential in the field of AI-powered science and research.
The Funded Projects
The 12 funded projects are diverse and innovative, with teams from universities and industry in the UK, US, and Europe. One of the winning teams, Lila Sciences, a US company, is building what it calls an AI NanoScientist, a system that will design and run experiments to discover the best ways to compose and process quantum dots, which are nanometer-scale semiconductor particles used in medical imaging, solar panels, and QLED TVs. Another team, from the University of Liverpool, UK, is building a robot chemist, which runs multiple experiments at once and uses a vision language model to help troubleshoot when the robot makes an error. Humanis AI, a startup based in London, is developing an AI scientist called ThetaWorld, which is using Large Language Models (LLMs) to design experiments to study the physical and chemical interactions that are important for the performance of batteries.
The Goals and Expectations of the Projects
Each of the 12 teams will receive around £500,000 to cover 9 months’ work, with the goal of demonstrating that their AI scientist was able to come up with novel findings. The teams are expected to design and run experiments, collect and analyze data, and document their findings in a way that can be reproduced and extended by others. As Rafa Gómez-Bombarelli at Lila Sciences notes, "The grant lets us design a real AI robotics loop around a focused scientific problem, generate evidence that it works, and document the playbook so others can reproduce and extend it." This emphasis on reproducibility and collaboration is a key aspect of the funding initiative, as it will help to advance the field of AI-powered science and research as a whole.
ARIA’s Experiment
The funding initiative is not just about supporting innovative projects, but also about taking the temperature of the field and determining how the way science is done is changing. Compared to the £5 million projects spanning 2-3 years that ARIA usually funds, £500,000 is small change. However, this was a deliberate decision, as ARIA wants to fund a range of projects for a short amount of time to see what works and what doesn’t. This will help the agency to determine the current state of the field and inform future funding decisions. As Rowstron notes, "To do things at the frontier we’ve got to know what the frontier is." By funding these projects, ARIA is taking a proactive approach to understanding the potential of AI-powered science and research.
The Challenges and Opportunities of AI-Powered Science
The development of AI-powered science and research is not without its challenges, however. As Rowstron acknowledges, there is a lot of hype surrounding AI, especially now that most of the top AI companies have teams focused on science. When results are shared by press release and not peer review, it can be hard to know what the technology can and can’t do. This highlights the need for rigorous testing and evaluation of AI-powered scientific systems, as well as transparency and collaboration between researchers and funding agencies. Despite these challenges, the opportunities presented by AI-powered science and research are significant, and ARIA’s funding initiative is an important step towards realizing these opportunities. By supporting innovative projects and taking a proactive approach to understanding the field, ARIA is helping to advance the development of AI-powered science and research, and paving the way for future breakthroughs and discoveries.