AI for Scientific Discovery

AI for Scientific Discovery provides an accessible introduction to the wide-ranging applications of artificial intelligence (AI) technologies in scientific research and discovery across the full breadth of scientific disciplines.

AI technologies support discovery science in multiple ways. They support literature management and synthesis, allowing the wealth of what has already been discovered and reported on to be integrated and easily accessed. They play a central role in data analysis and interpretation in the context of what is called ‘data science’.

AI is also helping to combat the reproducibility crisis in scientific research by underpinning the discovery process with AI-enabled standards and pipelines and supporting the management of large-scale data and knowledge resources so they can be shared and integrated, serving as a background ‘knowledge ecosystem’ into which new discoveries can be embedded.

However, there are limitations to what AI can achieve, and its outputs can be biased and confounded; thus, they should not be blindly trusted. The latest generation of hybrid and ‘human-in-the-loop’ AI technologies aim to balance human inputs and insights with the power of number-crunching and statistical inference at a massive scale, which AI technologies are best at.