3 ms·
Machine learning could fuel a reproducibility crisis in science
- gigama 4y ago"The high accuracy of predictions in tests often fools people into thinking the models are picking up on the 'true structure of the problem' in a human-like way..." A form of confirmation bias?