Yeah this seems like a decision yielded by a model. One of countless examples of the model’s outputs careening into individually defensible but holistically nonsensical results
replyA human seems less likely to embark upon a path that most humans know leads to something a human is unlikely to want
LLMs produce amazing things, but only one token at a time. Anything beyond that is retrospectively emergent at best
I guarantee the prompt “add auto update feature” was the first and last human input applied to that feature.
replydeleted
reply