No Digg Deeper questions have been answered for this story yet.
No Digg Deeper questions have been answered for this story yet.
I wrote something on Substack: The Prestige Flywheel - How AI’s "Dual-Hat" Culture is Bankrupting Our Intellectual Future https://milanfar.substack.com/p/the-prestige-flywheel
@docmilanfar One of the things I appreciated about ML coming from computational biology is that the culture of academic credit used to be much healthier than that found in bio, where 30+ people labs are not uncommon, and profs are more preoccupied with selling papers than scientific depth.
@docmilanfar Prestige Flywheel is just iterative greedy algorithm (in human) in action implication from optimization theory -local minima -1/e optimal even if submodular discrete-convex (still poor quality & 63% diminishing return in each iteration) human are just bad optimizer
@docmilanfar the prestige flywheel is a good frame for this tension
@docmilanfar I do think I know where it comes from 😁
Great piece, it was a real problem in our lab because my advisor did occasionally publish at NeurIPS, but was mostly trying to do high-prestige things in her subfield, like things that take time and work, papers that are worth reading ten years later. And very few of the students in the lab cared about that. They're like, why don't we do computer vision things? And she's like, in six months no one's going to read that paper, that's why. And I think she lost, honestly: the students did that and peaced out, one went to MSR after getting a master's: no interest in, you know, doing work for the ages.