When will an AI discover an important new result in Mathematics for the first time?
-
-
-
Replying to @haldaume3
"important new result". It hasn't yet happen, and we aren't anywhere close.
5 replies 0 retweets 0 likes -
Replying to @fchollet
eg which of these is not important and new:https://en.wikipedia.org/wiki/Computer-assisted_proof#List_of_theorems_proved_with_the_help_of_computer_programs …
2 replies 0 retweets 2 likes -
Replying to @haldaume3
theorem-proving != result discovery. Also these are done with a lot of (crucial) human help. Autonomous TP is crap right now.
1 reply 1 retweet 2 likes -
Replying to @fchollet @haldaume3
Autonomous TP as in, feed in axioms, a theorem, and let the prover run. Only very basic results can be proven that way.
2 replies 0 retweets 3 likes -
Replying to @fchollet
well, some were sufficiently unbasic that people couldn't prove them by themselves :) ... at least until Terry Tao came along ;)
1 reply 0 retweets 0 likes -
Replying to @haldaume3 @fchollet
unless there's falsifiable defn of "discv. imp. new result" I'm gonna stick with "been done" or "signpost will invariable move" :)
1 reply 0 retweets 2 likes
just improving autonomous TP to the point it can handle complex proofs (and new ones) would already be huge. We are far from it.
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.