Three pitfalls to avoid in machine learning

Researchers at TAE Technologies in California and at Google are using machine learning to optimize equipment that produces a high-energy plasma.Credit: Liz Kuball Machine learning is driving discovery across the sciences. Its powerful pattern finding and prediction tools are helping researchers in all fields -- from finding new ways to make molecules and spotting subtle signals in assays, to improving medical diagnoses and revealing fundamental particles. Yet, machine-learning tools can also turn up fool's gold -- false positives, blind alleys and mistakes. Many of the algorithms are so complicated that it is impossible to inspect all the parameters or to reason about exactly how the inputs have been manipulated. As these algorithms begin to be applied ever more widely, risks of misinterpretations, erroneous conclusions and wasted scientific effort will spiral.