> I was midway through explaining a concept he hadn’t covered when he stopped me. He pointed out that my way of speaking had completely changed and how it was unusually structured and didn’t give him the opportunity to ask follow up questions.
IMHO this sounds like a bit of an exaggeration in service of a specific narrative for the blog post, but language convergence has been a topic of conversation ever since the earliest autocomplete features appeared on smartphones.
The feedback loop in this situation is (LLM trains on people <-> People then train on LLMs).
Agree it's not common to give unprompted background context within any given normal conversation. People usually default to the pull style which Id agree is ultimately less efficient.
All that said, even though AI prompting is forcing the issue, which is good, the takeaway should be that _intentionality_ is very high leverage. Less so that it's because of some given (prompt) structure.
From the article:
> I was midway through explaining a concept he hadn’t covered when he stopped me. He pointed out that my way of speaking had completely changed and how it was unusually structured and didn’t give him the opportunity to ask follow up questions.
IMHO this sounds like a bit of an exaggeration in service of a specific narrative for the blog post, but language convergence has been a topic of conversation ever since the earliest autocomplete features appeared on smartphones.
The feedback loop in this situation is (LLM trains on people <-> People then train on LLMs).
Agree it's not common to give unprompted background context within any given normal conversation. People usually default to the pull style which Id agree is ultimately less efficient.
All that said, even though AI prompting is forcing the issue, which is good, the takeaway should be that _intentionality_ is very high leverage. Less so that it's because of some given (prompt) structure.