News for hackers. Stuff that matters.
Re:The Seven Deadly Sins of AI Predictions by yters on Sunday October 08, 2017 @04:14PM (#15429781)
Doesn't Chaitin's incompleteness theorem mean AI cannot learn? All axiomatic systems have a fairly low threshold past which they cannot identify random bitstrings, and all computational systems are axiomatic systems.
Re:The Seven Deadly Sins of AI Predictions by unlikelymordant on Sunday October 08, 2017 @05:06PM (#15429998)
If true, wouldn't that imply humans can't learn?
Re:The Seven Deadly Sins of AI Predictions by majewsky on Sunday October 08, 2017 @05:31PM (#15430107)
There is always the possibility that the human mind is more than the collective behavior of a bunch of neurons. It is conceivable that part of the human mind occupies some sort of astral plane like many people believe.

I'm not saying that it is. Science does not concern themselves with such theories (yet) because of Alder's Razor ("what cannot be settled by experiment is not worth debating") [1].

[1] https://en.m.wikipedia.org/wiki/Mike_Alder

Re:The Seven Deadly Sins of AI Predictions by yters on Sunday October 08, 2017 @05:43PM (#15430167)
If AI cannot learn and humans can, then this is experimental confirmation that humans are not AI, and the matter is settled. Therefore, the issue is worth debating.

I submit that AI cannot learn, per my previous comment. Furthermore, I submit that humans can learn, evidence being school, science and math. Ergo, humans are not AI.

Matter settled, debate over.

Re:The Seven Deadly Sins of AI Predictions by majewsky on Monday October 09, 2017 @09:44AM (#15433693)
Not at all.

1. Current AI is a miniscule subset of the set of all AI, so you need to argue why the observation generalizes to all possible AI.

2. The statement "AI cannot learn" is too ambiguous because you fail to define "learn".

Re:The Seven Deadly Sins of AI Predictions by yters on Monday October 09, 2017 @11:00AM (#15434320)
1. My statement applies to any computational approach to AI. If AI is possible it needs a halting oracle.

2. Learning is being able to recognize ever increasing levels of complexity. All axiomatic systems have a complexity recognition limit.

Re:The Seven Deadly Sins of AI Predictions by cassowary on Monday October 09, 2017 @12:44AM (#15431630)
If this is true it isn't apparent to me. You're stating it as if it's received wisdom. Could you clarify?
Re:The Seven Deadly Sins of AI Predictions by yters on Monday October 09, 2017 @01:02AM (#15431683)
It is not received wisdom. My argument is human learning entails recognizing an ever increasing amount of complexity, given infinite time, resources and motivation we can learn forever. But all axiomatic systems have a complexity limit, per Chaitin's incompleteness theorem, so cannot learn forever. Neither can a system add to its axioms. Therefore, human learning is beyond anything an axiomatic systems can do.
Re:The Seven Deadly Sins of AI Predictions by ScottBurson on Sunday October 08, 2017 @08:31PM (#15430805)
> all computational systems are axiomatic systems

Whatever gave you that idea?

Re:The Seven Deadly Sins of AI Predictions by yters on Sunday October 08, 2017 @09:31PM (#15431045)
They cannot be more capable of proving Kolmogorov complexity than an axiomatic system.
Re:The Seven Deadly Sins of AI Predictions by yters on Monday October 09, 2017 @12:09AM (#15431522)
Also the Curry-Howard correspondence.
Re:The Seven Deadly Sins of AI Predictions by red75prime on Monday October 09, 2017 @05:24AM (#15432455)
Citation from https://en.wikibooks.org/wiki/Haskell/The_Curry%E2%80%93Howa...

"[...] we can prove any theorem using Haskell types because every type is inhabited. Therefore, Haskell's type system actually corresponds to an inconsistent logic system."

Thus Haskell programs in general don't correspond to a consistent axiomatic system, which provides counterexample to your statement. Rebuttal complete.

Re:The Seven Deadly Sins of AI Predictions by yters on Monday October 09, 2017 @10:57AM (#15434291)
Inconsistent systems cannot prove Kolmogorov complexity limits. What's your point?
Re:The Seven Deadly Sins of AI Predictions by red75prime on Monday October 09, 2017 @01:24PM (#15435697)
Inconsistent axiomatic systems can prove anything. Thus either Haskell programs can prove the limits (and prove the opposite statement, as they are inconsistent) or they aren't axiomatic system as you mean it. Both possibilities contradict your statement:

> Chaitin's incompleteness theorem mean AI cannot learn

Re:The Seven Deadly Sins of AI Predictions by yters on Monday October 09, 2017 @01:37PM (#15435815)
If programs are not axiomatic systems, then they cannot learn, period. Only ones that have consistent axioms have a hope of learning, and even then it is severely limited.
Re:The Seven Deadly Sins of AI Predictions by yters on Monday October 09, 2017 @01:30PM (#15435745)
Inconsistent axioms can provide a valid proof of Kolmogorov complexity, but not a sound one. Thinking a false statement is true is not learning.
Re:The Seven Deadly Sins of AI Predictions by yters on Sunday October 08, 2017 @07:53PM (#15430677)
Down votes without counter argument means I'm right and you can't handle it.