LAUSR

Artificial intelligence (AI) and machine learning (ML) are increasingly ubiquitous in many areas of society, not least in medicine. AI offers the opportunity to harness computers to carry out more mundane tasks faster or better, for example more consistently, than humans. However, it also has the potential to move beyond what humans can do. For example, AI can combine many and disparate data sources and types into decision-making (e.g. genomics, imaging and pathology) or can interrogate very large sets of retrospective data to learn patterns where manual investigation would not be feasible or would not readily identify them, for example in early sub-clinical disease detection. For decisions and predictions, humans are restricted in the number of pieces of information that can be considered and processed simultaneously (typically said to be around 5), whilst AI systems are not so limited and can integrate complex data sets efficiently. Human decisions can also be influenced by bias and noise, that is extraneous factors causing variation in decisions between individuals or at different times. Structured decision-making can improve this, for example at a simplest level using checklists to aid the decision process, or codifying and extending these into algorithms and computerisation. Algorithms too can incorporate bias and must be carefully designed and tested to avoid this. Despite the potential of AI tools, they are just that, tools to supplement the clinical process and aid the clinician to be more