As I listen to discussions about the effects of LLMs on our industry, I hear a common refrain:
Sure LLMs have problems, but it's where the industry is headed.
I take issue with this statement because it is uncritical, ambiguous, and fatalistic.
The phrase conflates the wishful desires of company executives with the practical conclusions of software developers. When we say "the industry is headed" this way or that, do we mean developers are changing their tools? Or do we mean companies are pushing particular ways of working? I take issue with the ambiguity introduced by any analysis that omits this distinction.
No one knows what the future will look like (I made some predictions). But I find statements like the above overly confident. There's no room for nuance in such a sweeping statement. Saying "LLMs are here to stay" is factually true, but uninteresting. What matters is the degree of their use.
I could say more on the assumption that LLMs are somehow the objective, next stage of software engineering. I don't find this conception of technological innovation compelling. To me, LLMs are a clear result of our current Base/Superstructure dynamics. In another timeline they never would have been invented.