This sounds like a loaded question but... Why don't more people who talk about AI $subject (eg, ethics, policy, strategy) actually engage with foundational technical literature? What is the plausible justification for not habitually skimming Arxiv?
-
See also: https://twitter.com/kristineberth/status/997356628339904514 … It's really interesting, and I'll bet there's a good book in it.
An amusing test reporters could apply: "Oh, before we begin, would you mind quickly writing down the equations describing backprop in, say, a simple feedforward net? Okay if it's a bit rough!" AKA, do you know how any of this works?
-
-
Slightly unfair - there are people with interesting things to say who would fail this test. Still, I think the idea has merit, some kind of FizzBuzz for AI "experts".
-
I think I might fail this test, but I think I could explain, say, backprop in a way that made sense to them, then connect that to interesting innovations eg resnets, skip gates, etc, and tie this to progress in image recognition as a whole. (And that feels like it'd be useful)
- 2 more replies
New conversation -
-
-
This Tweet is unavailable.
-
This Tweet is unavailable.
-
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.