AI ain't that smart
Last week, there was a new article that got press for using AI to predict disease emergence. But as usual, it is buzz-word internet junk food and bad science. This is statistical modelling, and certainly worth-while, but nothing magical. "Actionable intelligence" = "a good guess"? We make guesses all the time about the science we study -- I don't understand why people get so excited about the possibility of computers doing the same. But a guess just a guess. Good scientific progress comes from doing the hard field and laboratory work to test a guess.
On top of that, our goal should be expanding human understanding, not replacing human understanding. An AI algorithm like this is basically relying on interpolation and naive extrapolation. Decades of study of biological systems has revealed that the details of living systems often matter -- Watson showed this in his beautify 1911 description of malaria prevention in Malay.
I like guessing and hypothesizing -- I'm a theoretical biologist, after all. But we should be promoting the men and women doing the hard work of science much more than those of us just crunching numbers. Talk is cheap.