Tired of that trope, I already wrote it before but "Those who can accept failure cheaply: firms that would otherwise hire interns, firms involved in rapid prototyping work, etc." is not correct.

I won't comment on hiring interns as that's not my expertise (even though if you want to teach your staff, obviously I can see a problem there) but I can comment on rapid prototyping, it's what I do.

Rapid prototyping is NOT making a CMS quick. It's not about making a quick mockup of a UI. It's not about making yet another well known... anything. The entire POINT of prototyping is to make something NEVER done before. Typically that means you are reaching the frontier. You are making something with NO documentation to rely on. You are using tools, hardware or software, which do NOT have tons of StackOverflow errors. There is no dataset to crawl, there is no well structured Q&A database to train on. You have to poke and see if the thing actually works as expected, and it often does not.

So sure, if you are using interns as a trick to underpay your staff, or if you are using prototyping as an excuse to build poor quality software fast, maybe it does help. If you are genuinely prototyping, it breaks fast and the supervision overhead makes it pretty pointless, especially since typically it's by actually implementing that you find out not just how the new setup works, but also its limits, and thus the actual needs of the project, not the one the stakeholder imagined would be.

So not, not for rapid prototyping either.

TL;DR: prototyping is a learning process, not a low fidelity output.

PS: this comes up very often from NON prototypists that I wrote a short piece about it https://fabien.benetou.fr/Content/GoodPrototypesAre10LinesLo... so much so that it feels like a pattern "GenAI/LLMs is good for tasks X" while the author actually does not do task X except very superficially.