Applied genetics. The mapping from gene to phenotype is not, shall we say, injective. Bring in the complexity of epigenetics, biochemistry and you have a comple

Applied genetics. The mapping from gene to phenotype is not, shall we say, injective. Bring in the complexity of epigenetics, biochemistry and you have a complex mess. Biological systems are tremendous black boxes with many second and third order effects that cannot be efficiently simulated and modelled (you can get pretty good approximations within a very narrow scope with supercomputers but those tend to have little practical use outside of basic scientific research). Another use case would be for pharmaceutical research, there are a number of companies already in the ML -> drug pipeline space but GPT has brought new insight to the table.