I agree that existing ML/AI systems focus on closed worlds. This is the fundamental reason that these systems are not safe to deploy in high-stakes open-world applications. But the idea that knowledge engineering will avoid these problems is puzzling. 1/https://www.nytimes.com/2018/05/18/opinion/artificial-intelligence-challenges.html …
-
Show this thread
-
Every hand-engineered system that I've seen suffers from exactly the same closed-world problem. Knowledge Engineering is also "Harder Than You Think". 2/
@GaryMarcus5 replies 3 retweets 36 likesShow this thread -
Replying to @tdietterich @GaryMarcus
Not the same, worse! One of the strengths of deep learning is that it is *less* brittle than approaches that relied on knowledge engineering. That said, the current mix of DL and knowledge systems seem to be able to cope with nearly open systems, like driving on real roads.
1 reply 0 retweets 1 like
except when they kill people or drive (2x) into parked fire trucks at 60 mph. but we i agree we need a mix. (& encourage you to apply same skepticism to cars that you do so well to radiology.)
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.