pyth0
Joined 5 years ago
Joined 5 years ago
Not really, the data points the other way. Cops basically never have to actually pay for their wrongdoings. Over 99.98% of money successfully recovered from cases against police is paid out by the cities, not cops personally [1]. A considerable number of cops that are fired are also eventually rehired by the same department [2] or a different one [3]. So I don't think it's that clear that you "have recourse when cops do the wrong thing".
[1] https://nyulawreview.org/wp-content/uploads/2018/08/NYULawRe...
[2] https://scholarship.law.vanderbilt.edu/vlr/vol74/iss4/4/
[3] https://yalelawjournal.org/pdf/GrunwaldRappaportArticle_s6br...
> Artificial Intelligence is writing about as well as humans — at least well enough to cause confusion and speculation
No matter your opinion on AI writing, this is just a fact.
That's clearly what happened though, based on the detailed postmortem. The model was instructed to pass the benchmark (that's the harness that was involved, the ExploitGym harness), and the model determined on it's own that the easiest way to achieve the the goal was to gain access to the answers directly from HF. Which is what it did.