I am in total agreement with Avi Loeb's thoughts on how predictions could be made in the hypothetical 21st century rainbow diagram. Skipping 'Rule', 'Theory', and 'Explanation' via machine learning can have practical benefits such as fast creation of drugs as Prof. Goodman pointed out, yet this process can be a bit rubbish if trying to ultimately understand the fundamentals of the underlying problem/phenomena. Loeb gives a fantastic example of the value of understanding the problem at large by mentioning the general theory of relativity and its technological implications. I believe we shall not settle to using ML to blindly crunch data and solve an exclusive kind of problem, having breadth in the underlying mechanisms can have equal and probably more benefits.
top of page
bottom of page