3 ms·
No one is claiming an agent can do 50% of arbitrary tasks. It's just 50% of METR's benchmark set. > I think you're overestimating, or oversimplifying Yeah if
by raincole 6mo ago
No one is claiming an agent can do 50% of arbitrary tasks. It's just 50% of METR's benchmark set.
> I think you're overestimating, or oversimplifying
Yeah if you only read comments on HN but not the actual linked article you will get oversimplified conclusion. Like, duh?
- boxedemp 6mo agoSorry for stating something so obvious. I'll comment less from now on.
- TeMPOraL 6mo ago> Yeah if you only read comments on HN but not the actual linked article you will get oversimplified conclusion. Like, duh? Curiously, for most submissions it's the opposite - comments are much more useful and nuanced than the source being discussed.