I'd rephrase it as "If a MTurker can log a task in 5s or less, we can probably automate it using AI". Includes the data-collection aspect
Ng's "If a person can do a mental task with less than 1s of thought, we can probably automate it using AI" is thought-provoking. Accurate?
-
-
-
It's likely that AI will make possible tomorrow any product that could be implemented today using MTurk (if cost was not a concern)
- Show replies
New conversation -
-
-
Accurate, but I don't like the use of "AI". NN or RL is better because it covers only the perceptual/reflex tasks, no deliberation.
-
'AI' continually gets redefined as "the stuff we can't quite do yet" Speech recognition, route planning, auto-focus, etc. were all AI once.
- Show replies
New conversation -
-
-
Hmm. If you model failure in 1s increments, an AI with 99.9615% accuracy would last 30 minutes, which is about the length of a game of Go...
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
So if we could break down any mental task to sub tasks taking less than 1s of thought, then we would be able to automate any task?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Better this quote of his than the other one about AI being the new electricity...
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Best to couch it in terms of flops/watt than just flops
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Yes if "probably"="more likely to". Otherwise fallacy: ML -> fast neq fast -> ML. Then again, ML neq AI
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.