DeepMind says reinforcement learning is ‘enough’ to reach general AI

Re:DeepMind says reinforcement learning is ‘enough’ to reach... by mjburgess on Thursday June 10, 2021 @02:18PM (#27463716)

It's a very interesting question.

Personally I take mammalian intelligence as the relevant standard we're actually aiming at.

So I'd say mouse+.

Houseflys, I think, are closer to non-intelligent than intelligent.

[ Parent | on HN ]

Re:DeepMind says reinforcement learning is ‘enough’ to reach... by dTal on Thursday June 10, 2021 @07:02PM (#27467087)

That's a little chauvinist! Birds regularly run circles around mice... er, so to speak.

[ Parent | on HN ]

Re:DeepMind says reinforcement learning is ‘enough’ to reach... by mjburgess on Friday June 11, 2021 @04:56AM (#27470862)

My view is: mammalian is sufficient, but not necessary.

Crow-level intelligence is probable likewise sufficient.

I think aiming at mammalian is a good long-term ambition. I think, either way, we are hundreds of years off.

[ Parent | on HN ]

Re:DeepMind says reinforcement learning is ‘enough’ to reach... by criddell on Thursday June 10, 2021 @04:06PM (#27465121)

Surely the AGI researchers have a benchmark though, don't they? Somebody else mentioned the Turing Test which is something...

[ Parent | on HN ]

Re:DeepMind says reinforcement learning is ‘enough’ to reach... by mjburgess on Thursday June 10, 2021 @04:32PM (#27465410)

I dont think there are any AGI researchers. At least, I dont think computer science has much to do with AGI.

The turing test is also not an AGI test, it's a "good enough" standard for fooling people.

Intelligence fundamentally requires a multitude of environmental capabilities. The turing test considers only a single i/o boundary.

[ Parent | on HN ]