A paper at the American Association for the
Advancement of Science meeting has questioned whether we are producing
erroneous science by the use of machine learning algorithms. The concern
being that these tools have been developed specifically to find interesting
things in datasets and so when they search through huge amounts of data they
will inevitably find a pattern. A lack of reproducibility in science may also be a result.
Others suggest that it is not the tools but
how they are used that produces problems.