Prediction: source code is going the way of assembly language

Prediction: Source code is going the way of assembly language.

20 years ago, people would look at you funny if you wrote “normal” code using assembler instead of a high level language (an exception that proves the rule: one of my most brilliant coworkers ever did write non-trivial windows programs using assembly language; ).

I think that soon the same will happen to the high level languages themselves. Another analogy: while you could program by editing .exe files, no-one would do that if they could avoid it – in the same way, most people will not look at source code any more.

The preferred format for “source code” might be something like 1) college textbook exercise level explanation of what is supposed to be done 2) a couple of complex end-to-end corner-case tests to make sure that everything is working right

Together, this “source code” and an AI model + “compiler” setup define a distribution of programs, and if the spec is tight enough, the vast majority of the programs in that distribution will be correct and doing the right thing. You can even run multiple of them in parallel.

Of course, there will still be a need to occasionally look at the source code - that’s also true for assembly in the modern era.

Thoughts?