90 comments 3 days ago: https://news.ycombinator.com/item?id=44596355

They’re probably aware. From the paper:

> We end by discussing how doubling-back aversion is distinct from established phenomena (e.g., the sunk-cost fallacy).

Is the author of this pop sci article aware? Hard to tell.

How is that relevant at all?

They pretend that behavioral psychology somehow isn't aware that perceived human irrationality is very common.

Finding just another example of that is not noticing a previously unobserved phenomenon.

Link should be https://arstechnica.com/ai/2025/07/exhausted-man-defeats-ai-...

Why not ask AI how to use <appropriate library / framework> to implement <thing>?

Doesn’t seem like you can blame NIH on AI more than other motivations for NIH.

Edit to add: If AI makes NIH easier, then it implies that AI is good at solving problems, and speaks to AI’s credit.

From me using Claude Code, without a proper system prompt, Claude generates code rather than using a library, this week e.g. command line parameter and flag parsing. The difficulty is where the tipping point is to use a library, it can't be https://www.npmjs.com/package/is-even

Dupe of https://news.ycombinator.com/item?id=44593357

This is post about a post. The original post was discussed here: https://news.ycombinator.com/item?id=44566996

Where did the term “system card” originate?

All I can find via google is that the term is used in Bridge (the card game) to coordinate play between partners.

First time I found this was in Google A2A protocol: https://developers.googleblog.com/en/a2a-a-new-era-of-agent-...

This is where model cards came from, as far as I know: https://arxiv.org/abs/1810.03993

239 comments on this post from yesterday: https://news.ycombinator.com/item?id=44585543

141 comments on post from ~8 hours ago: https://news.ycombinator.com/item?id=44591689

And 87 on this post, also from ~8 hours ago: https://news.ycombinator.com/item?id=44591785

my bad