“We’re using the phrase artificial intelligence increasingly, but there’s very little intelligence in what’s happening. It’s just learning. And there are many issues still be faced in learning itself.”
Keith Baggerly, Professor of Bioinformatics and Computational Biology at the University of Texas MD Anderson Cancer Research Center, can be thought of as a data detective. Bioinformatics involves assessing high-throughput molecular diagnostics, typically thousands of measures of genes at a time, for suitable medical treatments.
But Baggerly’s work has often centred on attempts to reproduce work previously done by others, which is typically hard to do in this field due to a lack of methodological precision in the analysis of large datasets. Baggerly emphasises the importance of clean data for improving accuracy. He also advocates the pre-specification of hypotheses and models to be used in an experiment in order to guard against false positives or negatives.