Machine-learning techniques used by scientists are leading to inaccurate or irreproducible results. We must always be mindful that we are the easiest person to fool and that we don't allow these new computing powers to simply supercharge our biases. https://www.bbc.com/news/science-environment-47267081 …
And, machine learning systems can't explain *why* they do what they do. These "down sides" to machine learning are well known to anyone who understands machine learning. We've known this since at least the 80's when I first studied machine learning. Only laymen expect more.