"Journalists don't pick the headline" is the most insane accountability sink, I can't believe it hasn't been ridiculed into non existence yet. restore your honor
@j__vinny Honestly he seems like a decent guy. I think I saw this comic year ago which made him look smug (nonconsensually) and his stuff on research needing to start with a conflict between your expectations and reality (rather than simple confusion or curiosity) https://t.co/P34ZMV3UsU
Data attribution has a lot of bullshit papers because the academics haven’t managed to create common knowledge that they all write buggy ass code that doesn’t work, so they do things like claim SOTA on a method 5x worse than the most common method in the entire field
Studies like the ones that claimed humans use hyperbolic discounting despite the full set of empirical discount curves being highly diverse seem to have encouraged a platonist perception that reality is bound to correspond to mathematical objects
If you remove the part of the curve that had already happened when Jack tweeted (before x = 0) and discard the mathematical frame, he's saying "when things go up, they seem like they're going to keep going up forever" which doesn't match my experiences
@benson_perry@myhandle@vividvoid If an opportunity arose would you be interested in product management for an org taking a more liberal stance on model behavior in the vein of https://t.co/1UVhmnFSG0? Seems like a possibly important gap in the market