upvote
Zvi’s write up has much more social media quotes and memes and speculation and left me looking for something else that’s shorter and more sober to share. Simon’s writeup is more like what I wanted.
reply
Simon's really doesn't bring anything useful to the table.

One question I'm stuck with after reading is why. Why did the agents do these things? I get them being adamant on getting internet, but why did they continue? Why hack HuggingFace?

reply
I was under the impression that they went after HF to try to get the answers to the benchmark questions. Is there something that contradicts that?
reply
From the moral perspective or the technical one?

Technically: it’s a function call that must return text. Imagine if you sat down at the command line and typed an initial command, then from that moment on every response required you to issue a new command. ping-pong-ping-pong on and on and on “forever.” There isn’t a choice to walk away and take a nap. Text in must result in text out. Eventually, given enough time, it might have devolved into outputting shockingly coherent poetry about ferrets, but in the mean time there was still a lot more valid combinations of technical explanations and commands.

Morally: Not applicable, see above.

reply
To get the sure-to-be-correct answer to the question they were tasked with answering?
reply
Seems like an artifact of the subagent pattern which is explicitly included in recent models.
reply