Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

venturebeat.com

Post reply on HN
Print view View original

41–50 of 181 posts

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#41

Post by tanilama »

skybrian wrote: post #29
karmasimida wrote: post #17

And no one should be surprised by this. The NN advancement of late doesn't help addressing human-style symbolic reasoning at all. All we have is a much more powerful function approximator with a drastic increased capacity (very deep networks with billions of parameters) and scalable training scheme (SGD and its variants). Such architecture works great for differentiable data, such's images/audios, but the improvement…

Hmm, it seems like natural language translation has been getting quite a bit better with statistical techniques, though? I guess it depends what you mean by "only incremental".

I still remember sometime back Google's GNMT translate the Chinese text of 'I don't want to go to work' into English 'I want to go to work'. That example alone should be sufficient to showcase how most advanced machine learning model can fail at the simplest task.

It didn't understand the source material, it is just very good at memorizing and faking.

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#42

Post by m0zg »

skybrian wrote: post #29
karmasimida wrote: post #17

And no one should be surprised by this. The NN advancement of late doesn't help addressing human-style symbolic reasoning at all. All we have is a much more powerful function approximator with a drastic increased capacity (very deep networks with billions of parameters) and scalable training scheme (SGD and its variants). Such architecture works great for differentiable data, such's images/audios, but the improvement…

Hmm, it seems like natural language translation has been getting quite a bit better with statistical techniques, though? I guess it depends what you mean by "only incremental".

It's been "statistical" since about mid-00's. What's new is that it's now neural.

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#43

Post by anonytrary »

roenxi wrote: post #30

The computational power of the hardware is getting really close to what a human brain is capable of (on an exponential scale, anyway). If "nowhere close" means not in the next 5 years then sure. Over the medium term I'm not sure AI researchers are the best people to ask. They are completely dependent on how much power the electrical engineers give them - I doubt there is a deeper understanding what a doubling or quad…

> The computational power of the hardware is getting really close to what a human brain is capable of

Why do you say that? AFAIK computing architecture and brain architecture are completely different. How would you even begin to compare their power?

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#44

Post by codekilla »

WhompingWindows wrote: post #37
codekilla wrote: post #27

Earlier quoted context omitted.

> Behind every successful neural network is a human brain. I've spent a lot of time trying to explain this to people, that there is a confluence between the human brain and the machine, people tend to look at the machine separately, which is a mistake. When I say unequivocally, 'there is no such thing as machine intelligence', I just get blank stares.

Arguably, there are successful brains behind every successful brain, too. Every great innovator and thinker was building off the backs of numerous other thinkers and teachers in their life. Should we be surprised that it's much easier for a tool+human(s) to do better than a tool alone, given we also expect a single human + human(s) as colleagues to do much better? Never mind the whole learning/development process, du…

Sure, I'd agree. But this brings up the idea of autopoiesis, and then I think things get really murky.

One question that interests me is this: Does intelligence have as a prerequisite a living system, such as a cell? If so, what is our definition of the living system and why is that important? If not, what abstract qualities of intelligence are we really trying to capture?

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#45

Post by darkmighty »

odyssey7 wrote: post #20
karmasimida wrote: post #17

And no one should be surprised by this. The NN advancement of late doesn't help addressing human-style symbolic reasoning at all. All we have is a much more powerful function approximator with a drastic increased capacity (very deep networks with billions of parameters) and scalable training scheme (SGD and its variants). Such architecture works great for differentiable data, such's images/audios, but the improvement…

Are there ways that an AI practitioner would be able to tell whether a neural network is doing human-style symbolic reasoning?

You'd be able to tell if every brain structure was replicated by a NN analogue (and we understood them sufficiently well). Otherwise you can only use behavioral replication (i.e. Turing tests) to infer it.

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#46

Post by joe_the_user »

sdenton4 wrote: post #36
karmasimida wrote: post #17

And no one should be surprised by this. The NN advancement of late doesn't help addressing human-style symbolic reasoning at all. All we have is a much more powerful function approximator with a drastic increased capacity (very deep networks with billions of parameters) and scalable training scheme (SGD and its variants). Such architecture works great for differentiable data, such's images/audios, but the improvement…

Maaaaybe. I tend to think that symbolic reasoning is a learning tool, rather than a goalpost for general intelligence. For example, we use symbolic reasoning quite extensively when learning to read a new language, but once fluent can rely on something closer to raw processing - no more reading and sounding out character sequences. Similarly with chess - eventually we have good mnemonics for what make good plays, and…

Once someone is fluent in a language, the logical operations and judgements involved stop being overt and highly visible to the conscious mind. But that doesn't mean that one stops getting the benefits and results of logical operations.

What you might see as logical operations "not mattering", I would see as logical operations integrated so deeply into reflexive operations that it's hard to see where one ends and the other begins. The contrast is that humans can do pattern recognition in a neural net fashion, taking something like the multidimensional average of a set of things. But a human can also receive a language-level input that some characteristic is or isn't important for recognizing a given thing and incorporate that input into their broad-average concepts. That kind of thing can't be done by deep learning currently - well, not a non-kludgey sort of way.

Similarly, I have a soft-spot for the view that a mind is only as good as its set of inputs.

It depends on how you want to mean that. A human can take inputs on one thing and apply them seamlessly to another thing. Neural nets tend to be very dependent on the task-focused content fed them.

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#47

Post by colordrops »

siekmanj wrote: post #7
yters wrote: post #6

As I always ask regarding this sort of story, why do we believe human intelligence is computable? The only answer I've heard is the materialist presupposition and sneers at any other metaphysic as "magic," which is not exactly a valid form of argument. As an alternative, the human mind could be some sort of halting oracle. That's a well defined entity in computer science which cannot be reduced to Turing computation,…

We know the brain is doing something - if you don't want to call it computation, then you might as well call it magic.

Are you positing that the only alternatives are computation and magic?

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#48

Post by mannykannot »

sjeohp wrote: post #40
yters wrote: post #19

Earlier quoted context omitted.

Well, quantum computation is weaker than a nondeterministic Turing machine, so not the same thing I'm saying. Penrose correctly identifies the mind cannot be a deterministic Turing machine, but his invocation of quantum mechanics does not solve the problem he points out. A DTM can simulate an NTM and hence anything inbetween, so the inbetween of quantum computation does not solve anything. The fundamental problem Pen…

>Hence, a halting oracle is the best explanation for the human mind What does it explain though? That the human brain has a black box capable of solving certain problems... how exactly?

Indeed - it's essentially the homunculus fallacy, or magic dressed up in the language of knowledge.

https://en.wikipedia.org/wiki/Homunculus_argument

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#49

Post by brandonmenc »

anonytrary wrote: post #23

Tangential: This title is weird. As if no one but the top minds in AI didn't know this? This isn't big news to anyone who has done even just a modicum of AI research.

> As if no one but the top minds in AI didn't know this?

Anecdotal, but nearly all of my programmer friends believe that full-blown AGI is less than a decade away.

Re: Geoffrey Hinton and Demis Hassabis: AGI is nowhere close to being a reality

#50

Post by thanatropism »

I don't believe in the idea of AGI for Dreyfusard reasons, but it's possible that it could emerge from something completely different than deep learning.

For all we know, Isabelle and Coq could be speeding through the road to consciousness but we're busy having a blast doing Computer Vision pretending it's AI.

Post reply on HN