CS professor literally has a mention of the halting problem up on the slides and just said ML could maybe automate debugging...

@AOL ive just started to internally replace ML with "magic" when i read it

@dankwraith This is the most boggling one to me because it's provably impossible and he knows it! I feel like tech peoples brains just turn off for a bit when they talk about ML

@AOL people think ML can escape from the turing machine box somehow

@dankwraith @AOL curious what will come after ML when the hype finally dies down. or is it just gonna come back again forever?

@AOL I mean, you can use a neural net to tune a fuzzer, I guess

@AOL I mean, the halting problem also demonstrates that humans aren't capable of debugging, right? It may be the case that ML will never match human ability in debugging, but the halting problem doesn't show that.*

*(I'm taking it as an article of faith that physics, and therefore human behavior, is computable. If we ever find any non-computable physics it'll be fascinating, but for me it's in the same category as polytime algorithms for NP-hard problems: "ignore 'em until we find 'em".)

Sign in to participate in the conversation
Radical Town

A cool and chill place for cool and chill people.