Three road blocks to solve for artifical general intelligence (#AGI)
1) Massive Multitask learning with a single joint model
2) Ability of algorithms to update their objective functions in continuous learning
3) Learnable combination of fuzzy/fluid and symbolic reasoning
-
My main concern is treating "abduction" as some special kind of inference. Probabilistic reasoning unifies deduction, induction, and abduction in a single, elegant framework. The difference isn't the kind of reasoning but the type of objects over which we are reasoning.
Maybe, but how then to make all that work in, say, making the kinds of inferences that Sherlock Holmes made or in devising a machine that could induce a double helix from the data Watson and Crick had available. @yudapearl is right that causality is a key part of the mix.
-
-
Abduction need not be causal, but if it *is* causal, there are many benefits, of course!
-
The analogy one should make between induction, deduction and abduction is how it relates to available training data. It mirrors interpolation, extrapolation and inventiveness. See: https://medium.com/intuitionmachine/the-key-to-general-intelligence-is-the-mechanism-of-artificial-ingenuity-dad26bc99043 … .
- 6 more replies
New conversation -
-
-
Agreed causality is key part of the mix, to make it work there are 2 whys. Why in the beginning is intentional, in scientific procedure this is the purpose, 2nd why is a quest for information about the conclusion. Must know first why to determine relevance of the 2nd.
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.