A thoughtful take from @bcantrill and the folks at @oxidecomputer on the use of #LLMs in our profession.
This part in particular examisnes a key tension: "LLM-generated prose undermines a social contract of sorts: absent LLMs, it is presumed that of the reader and the writer, it is the writer that has undertaken the greater intellectual exertion. (That is, it is more work to write than to read!) For the reader this is important: should they struggle with an idea, they can reasonably assume that the writer themselves understands it — and it is the least a reader can do to labor to make sense of it."
"If, however, prose is LLM-generated, this social contract becomes ripped up: a reader cannot assume that the writer understands their ideas because they might not so much have read the product of the LLM that they tasked to write it."
@igb And it's not just prose - it extends to code and art the same way, even if it's a bit less obvious there.
@dos @igb i feel like with a tool I could excuse it if it worked well - if someone used say a genetic algorithm to design the ideal wrench that actually produces an ideal wrench that to me wouldn't feel any less valuable than a manually-designed wrench
but with anything artistic or creative the assumption is that someone intentionally created an experience for you to experience in a given way
@zuthal @igb However, programming is, for the most part, a social activity. You write code not just for machines to execute, but also for humans to review and maintain, and until that changes, the intention behind the structure of the code you output is just as important. A big part of my job is to read, understand and fix code written by others, and there's a similar assumption of "someone who wrote this at least believed that it would make sense" there.
