excerpt from discussion with a friend who is writing about AI/MLpic.twitter.com/jC9rUjTnpR
Show this thread
-
if anyone has references to any philosophical or critical discussion of this argument about “understanding” i’d love to read more, this is mostly what i’ve noticed from following research at a distance
Show this thread
This is not a philosophical argument, what is meant is simply "when a human solves problem A (e.g. image captioning), the corresponding ML algorithm solves problem B (the problem of mapping inputs to outputs over the training data), which has only very superficial overlap with A"
-
-
Similarly "they just do statistics" simply means "DL models rely on surface statistical characteristics" (lots of good papers about this) -- it's a statement on generalization.
-
i think “what does understanding mean?” is a philosophical question. are you saying: “understanding” implies an ability to generalize in a way that exceeds the capability of (or is somehow different from) our models right now?
- Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.