Twist: what if intelligence was actually the ability to *set* behavior goals, not just achieve pre-set goals?https://twitter.com/Numenta/status/730950722959740928 …
I think long-term planning and goal-setting are fundamental problems in AI, that almost nobody is trying to solve.
-
-
According to some, the algo of intelligence is a LSTM, according to others it's HTM, etc. But short-term sequence prediction != intelligence
-
Scaling up sequence prediction, or generally scaling up the ability to map a sensorimotor space & environment, doesn't give you intelligence
- Show replies
New conversation -
-
-
Fully agree - but "dynamic" reward functions only make sense in a social setting
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Would you consider loop closure - revisiting a place and able to tell you've seen this before, for robot navigation for long-term?
-
right now solved as image retrieval but perceptual aliasing and changing weathers, and viewpoints make it very hard..
End of conversation
New conversation -
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.