@johnpaulflintoff @pluralistic You only use an AI summary if you don't really care about the answer. Which begs the question, why get an LLM to summarise anything?

If you need to make an informed decision about something, you need to read it and understand it yourself.

You run the risk of making decisions based on hallucinations, with potentially catastrophic consequences.

@arclight On its face, the quoted argument seems to suggest that I can't improve my understanding of a subject by reading a human-written summary either. But sometimes reading a human-written summary does improve my understanding.

Does the linked article address this objection? If so I'll be interested to read it, but the quoted sentence doesn't seem promising.

@johnpaulflintoff @pluralistic Learning to read critically and question your assumptions and the claims of others are things you learn in those liberal terrorist training camps they call "college".

Is the US you will get very little of that in the occupational training you receive as a child.

To see that you need to understand enough to know how to evaluate a claim you have to first realize you should evaluate the claim before just believing it.

@johnpaulflintoff @pluralistic

This right here is my chief complaint with the whole "AI" "assistance" proposition. If I have to check its work thoroughly, how is that an improvement over just doing the work from the start? Even when dealing with other humans, unless the other person is really on their game, I nearly always find it easer/faster to do it over from scratch myself. I can't seen any angle from which "AI" looks like an improvement.