@karpathy@plasticlistorg Food and health is one of the few domains where regulation and standards help. Because short term companies do not care about your health, they just care that you buy their product
@yacineMTB It's also about opportunity cost. Doing Y can imply not doing X if you assume finite amount of time and resources. So gotta be strategic and then people can be caught in decision paralysis
@VillageWest14 A hedge fund isn't there to make you as an LP more money than the SP500, rather than make you feel like you've put in agency to do better (psychologically). Also some funds do actually have better beta, which isn't really covered by just comparing annual returns.
@kimmonismus Imagine an Autonomous Vehicle company telling you they succeeded on ImageNet or some nice 3D shape recon eval. There will be no academic benchmark that can convince me of a working AV product besides the actual product working. It surely works for theoretical fields like math tho
@kimmonismus I wonder why LLMs became so benchmark heavy. In computer vision/robotics, you aim to solve a concrete real world problem, and derive useful proxy benchmarks along the way (internally). I don't need benchmarks to feel the AGI, I'll feel it if it works and gives me tangible value.
Like most things in life, Project Mariner only happened thanks to a legendary team of dreamers who were willing to work through the stack to make Gemini truly agentic and capable of collaborating with computers in the same way as humans! <3
@elonmusk There needs to be more holistic coverage though. I wouldn't have heard about what happened in Syria the last few days solely based on my X feed.