While I agree with the overall drift of this article, neither intelligence nor common sense are all-or-none capabilities. Ferrucci says DL systems have "no understanding"--WRONG. They have extremely shallow understanding. 1/
“If we have @ElonMusk and Nick Bostrom talking about ‘superintelligence’, we need [sceptics like] @GaryMarcus to provide a reality check.” - @etzioni, CEO @allenai_org, in excellent new @FT piece on danger of overestimating AI and #deeplearninghttps://www.ft.com/content/4367e34e-db72-11e7-9504-59efdb70e12f …
-
-
-
Etzioni says computers need Common Sense. Better to say that they need a lot more common sense than they have now. There will be no moment when computers "have" common sense, just as we people continually reveal holes in our common sense.
- 1 more reply
New conversation -
-
-
I completely agree. As a law professor who works in the artificial intelligence law and policy space, my work is made much harder by the exaggeration of current AI technology and the (unsupported by evidence) but confident speculations about the future of AI.
-
I always use a phrase from Harvey Brooks: science should be the conscience of technology.
End of conversation
New conversation -
-
-
Add
@SamHarrisOrg to that list. It's easy to philosophize about AI, try building it. There's a reason experts in the field aren't sounding alarms. Highly recommend the podcast though... he should have@GaryMarcus on!Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
This Tweet is unavailable.
-
-
I usually use the analogy 'getting closer to the Moon by climbing a tree'. Not a known speed-of-light type limit, but no path from AI research to date to human abstraction or reasoning. Even in wildest scientific research there's some plausible hypothetical path one can point to
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Shoutout to
@beenwrekt for also helping keep it real.Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
This is odd because we just *dont know* how far we are from having human-style generalizable AI, because we don’t know how we work. We might be very far, or very close. It seems hard to deny that there is *an* answer, and we are getting closer, increasingly rapidly.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I have yet to read a skeptics view of the dangers of an AGI be other than the defense that "we are not there yet". That argument doesn't nullify the need to consider the implications of an AGI as AI gets less and less narrow.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
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.