Very few people write assembler code today, because compilers are good enough most of the time.
The same can happen if we get generic and robust enough ML tool.
Could you expand on that analogy? I'm not sure if a one-model fits-all is comparable to a one-optimizer-fits-all. Wouldn't that be an argument against transfer learning, since we don't have a universal programming language and thus there does exist multiple compilers?
Isn't it a big difference that people do understand assembler in the small?