I do have my doubts that plain neural networks will ever be able to achieve conceptual understanding.
I have an affinity for classical, rational AI in that you can correct it and it will take that correction and instantly apply it to is knowledge base. It can also explain why it came to a conclusion. (though obviously this style has its very real limitations)
NNs and other current statistical/connectionist approaches don't really have this capability, which I see as a necessary part of human-level intelligence. They are trained to "get a feel for" particular inputs to indicate particular outputs. If you were to personify the NN analyzing the dog/"ostrich" picture, and ask it why it thinks it's an ostrich, it could only reply with "I dunno, it feels like it's an ostrich". The only way to correct it is to retrain it with careful checks so that it behaviorally "gains a sense" of what looks like a dog vs ostrich more reliably.
Many language/word based features operate similarly. Watson, Siri, Google search, all sort of map strengths of relations with your input words & patterns to some associated results that it just sort of statistically was reinforced with. These can yield information that is of real use for a human to further evaluate and act on, but I wouldn't trust such a system to directly act on those associations; they're wrong too often.
But there is no possibility of actual conceptual discourse with NNs as we know them, to correct them, to inform them, or to ask them to explain their results. This is a fundamental barrier to achieving human-style intelligence.
This is not to say that there aren't NN-based possibilities that might work, like tacking together multiple interacting NNs, each which have the possibility to specialize on concepts and influence other NNs. But too much has focused on single NNs and direct input-to-output "monkey-see, monkey-do" training. It's a manifestation of the Chinese room problem.
At the risk of nitpicking, I'm going to specifically address your statement of "It feels like an ostrich"
To my mind, that's EXACTLY how humans do it. A baby is not instructed, "it has 2 legs, feathers, and is tall, therefore it is an ostrich", there's quite a bit more "LOOK OSTRICH /present input of ostrich/" prior to the point of being able to generate any justification.
As sister posts have pointed out, I tend to believe that the ability to justify the classification is another learned skillset that comes later developmentally, the ability for higher primates (and babies) to perform classification without the (as far as I know) ability to do the higher order reasoning lends itself to this theory.
I find how my two-year old classifies birds to be amusing:
1) He called all birds ducks for the longest time, presumably based on being exposed to a bath duck, and this weird Youtube video about ducks swimming in the water.
2) Then he called some of them bird, based on being exposed to a bath toy bird.
3) Then he split them up a bit more... parrot, chicken, rooster, etc. He started seeing them in books and real life.
4) At no point could he explain the difference. Still can't.
5) And he definitely thinks the only difference between a rooster and a chicken is colorful feathers.
6) And he's not quite sure about ducks and geese, but usually gets it right.
I guess he still needs to develop the advanced bullshitting (sorry, I meant "rationalisation") skills required to say, "Well, I think it's a duck, because it looks like my internal picture of what ducks are supposed to look like, dear sir!"
I think one of the major differences is that humans use multiple independent algorithms when it comes to deciding if there's an ostrich in the picture. And most of these are far more sophisticated than the NNs. So in practice the way we solve classification problems is fundamentally different, even though it may seem/feel somewhat similar.
Most prominently there's object recognition (telling things apart from the background, recognizing parts of objects, etc.), size/distance determination (which uses over a dozen separate algorithms: https://en.wikipedia.org/wiki/Depth_perception). We can also reconstruct a 3d model from a 2d image and compare that to our idea of what ostriches look like in 3 dimensions.
So even if it is done unconsciously, we recognize an ostrich because (among other things) it's the size of an ostrich. The NN has no idea what size is, let alone the size of an ostrich relative to other things.
I think the next big breakthrough in machine learning is in ensembles containing NNs trained for completely different things, but acting in a complementary fashion.
I agree that the initial human classifier is the "general feel". My distinguishing point is in the retraining. We can form conceptual constructions of what we see to resolve training correction, and once that is bootstrapped, carry the formed concepts and the meta-cognitive capability of concept creation over into new contexts.
If we've only seen automobiles, then all of a sudden are introduced to trains, 18-wheelers, motorcycles, etc, we are challenged to conceptually organize and discover the differences between these classifications. There's still a fair amount of "feel" in weird overlap cases, but the conceptual distinction is created after the fact. There is no mechanism for that in NNs.
Using the word "feels" (or "thinks") is a too much of a personification of a neural network. You cannot forget that under the hood, it is cranking through a bunch of math functions. You always get the same output from the same input. That function is not going to change on its own, and it is not going to suddenly gain new abilities to express itself in words. It is not a human, so there is no point in assuming it will behave or grow like a human.
I do have my doubts that plain neural networks will ever be able to achieve conceptual understanding.
I have an affinity for classical, rational AI in that you can correct it and it will take that correction and instantly apply it to is knowledge base.
I don't claim this to be any amazing insight, but I strongly suspect that the human brain works on some combination of both probabilistic methods and something like the symbolic logic of GOFAI. How many "systems" there are, and how they interact, is an open question, but I really do think there's "some there, there".
Which reminds me, I need to get back to reading "Thinking, Fast and Slow", which I started a while back and got distracted away from.
I'm a firm believer that it's a mistake to ask machines to do X in the hope that they'll do Y. We can't train an ANN to classify images then ask it to explain its reasoning; that's not the task we trained it for!
How might we train an ANN to explain its reasoning? One approach would be to learn programs: have the ANN write programs which classify images. Then we have a classifier (run the program) and an explanation of how it works (read the program). We don't have an explanation of how the ANN chose the program, but again, we didn't train it to tell us. In principle, we could keep adding meta-layers; in practice, the search space and evaluation time would explode :)
Yes, exactly. We're not training consciousness networks. That's not even a goal of all this research. We're training dumb high-speed classifiers.
How might we train an ANN to explain its reasoning?
There are networks that exist today that'll happily explain an entire scene to you (the whole "this picture contains a pizza sitting on an oven in a kitchen and there is a dog in the corner"), which is closer, but still just pairwise training. The "thought" process isn't recurrent or turing complete so it can't make progress on its own.
> There are networks that exist today that'll happily explain an entire scene to you (the whole "this picture contains a pizza sitting on an oven in a kitchen and there is a dog in the corner")
I was thinking more along the lines of: "This picture contains pizza. This is because there is an ellipse which appears to be covered with cheese and pepperoni. I say that because the dominant colour is yellow and there are elliptical patches of a more reddish colour. ..." and so on, down to the pixel level.
We can kind of get this by running the networks backwards, but it's not really deductive reasoning; for example, we can't correct the network by saying "that yellow ellipse covered with red ellipses is actually a pile of vomit", and watch that knowledge propagate through the weights. Instead, we have to re-train with more examples of pizzas and vomit.
> There are networks that exist today that'll happily explain an entire scene to you (the whole "this picture contains a pizza sitting on an oven in a kitchen and there is a dog in the corner"), which is closer, but still just pairwise training. The "thought" process isn't recurrent or turing complete so it can't make progress on its own.
Aren't those generated by RNNs, which are Turing complete?
Even humans can’t explain their own reasoning. We can only rationalize our intuitive decisions after the fact, but it’s pure hokum – all decisions are made intuitively using no rational thought whatsoever, only afterward can we make up rationalizations for it.
It follows that it should be perfectly possible for an A.I. to do this, too.
(Edit: I now realize that this rationalization process was theorized by Douglas Adams in Dirk Gently’s Holistic Detective Agency, as described by Wikipedia:
“The story also mocks the corporate world with [the software] package called Reason, which inverts the idea of a decision-making program. Instead of proceeding from ideas and logic to a decision, it takes a decision that has already been made and creates a reasoned justification for it.”
Let's say you want as much money as you can get. This is probably not rational. Now let's say you have a choice of being given $5 or $10, which do you choose? Why was that not a rational choice? Because it's predicated on an irrational desire?
It's trivial to brute-force an answer to your question: if I take $5 I'll get $5, if I take $10 I'll get $10. I prefer $10 to $5, so I take the $10.
Intelligence is needed when there are too many possibilities to brute-force. We need to make guesses about which actions may give better results; we have to spot patterns, generalisations and simplifications to allow re-use of previous experience; we need to identify the crucial aspects, in order to narrow down the possibilities; we need to allocate resources efficiently, and know when to give up; and so on.
These are the things we would like machines to do well, but we have no satisfactory theory to explain, measure or compare such things. Introspecting our own thought patterns is not an effective way to reverse-engineer these proceses in humans, as the parent says.
No, that is not what I meant – the choice itself is not made rationally, even though a theoretical rational actor might have made the same choice.
The most we humans can do it to make an intuitive choice, allow the post-hoc rationalization to occur and then check its rationality by analyzing its logic. If the logic is bad, we go back and try another intuitive choice.
By your own logic, you just made an intuitive decision about whether I have a point, and now you're just rationalizing it. Not much point in trying to have a rational conversation.
Is your belief falsifiable? Wouldn't the experimenter just be rationalizing their intuitive decision about the meaning of their observations regardless?
The biggest problem I see to training a network to program is that it would require a lot of example programs to serve as training data. Which means we need a lot of really good, bug free programs that already accomplish similar tasks. And in my opinion, we do not have enough reliable seed programs to make anything useful, at least currently.
It is very easy to take a picture of a car and tell a computer what it is, not so easy to write a program and explain to a computer what it does.
IANANR (I Am Not A Neuro-Researcher), but it seems to me the "only" step you're requiring is that of self-consciousness. We can explain why we think it's an ostrich because we can point our neural networks at themselves, and say, "what sort of thinking process does this look like? Oh, it looks like visual pattern recognition of certain features like a long hairless neck, bird features, etc, so that's why I think it's an ostrich."
We are capable of conceptual thinking because we can think about our own thoughts, and further think about our own thoughts about our own thoughts, etc, in an endless cascade of meta-thinking (which is in fact what thinking is). The ability to experience a gestalt, and the ability to think about it, are two major steps. We seem to have achieved the first, with current NNs. That's pretty amazing already!
I think the ability to be pointed at itself is a necessary step for those NNs to develop self-consciousness and, thereby, conceptual understanding. We're not there yet, for sure, but how much further are we? A few years? A decade? A few decades? The latter is no doubt a very pessimistic estimate (conservative AI researchers agree on this).
Human-style intelligence is certainly not here yet. But it is almost certainly just around the corner.
To clarify, I'm not saying human-like AI is impossible or unreasonably far off.
The author seems to be implying that what current NNs show is a viable form of intelligence that just happens to "think" in a different way that we need to figure out.
I'm saying this particular manifestation of trained behavior is not viable for expanding into human-level intelligence, as there's no hope of meta-cognition (I don't use the term "self-consciousness" purely because it's overloaded with unreasonable fields). NNs would have to fundamentally change or be replaced in order to gain such. When we do achieve machine conceptual cognition & meta-cognition, it will be a system that will not "think" like feed-forward trained NNs, even if it shares some history or composition with that architecture.
>>> Neural net ... "I dunno, it feels like it's an ostrich"
I don't think this applies to deep-learning. Because with deep learning , the lower layers will will say "this has long neck", "this has legs" , etc... and that will help the higher layers to to understand "it's an ostrich".
There are many specialized areas of the human brain.
It's not a single, unidirectional, backpropagated, simple-coefficient, neural net. It is not a recurrent neural network, nor is it a single instance of any particular NN-derived or NN-ish network.
I have an affinity for classical, rational AI in that you can correct it and it will take that correction and instantly apply it to is knowledge base. It can also explain why it came to a conclusion. (though obviously this style has its very real limitations)
NNs and other current statistical/connectionist approaches don't really have this capability, which I see as a necessary part of human-level intelligence. They are trained to "get a feel for" particular inputs to indicate particular outputs. If you were to personify the NN analyzing the dog/"ostrich" picture, and ask it why it thinks it's an ostrich, it could only reply with "I dunno, it feels like it's an ostrich". The only way to correct it is to retrain it with careful checks so that it behaviorally "gains a sense" of what looks like a dog vs ostrich more reliably.
Many language/word based features operate similarly. Watson, Siri, Google search, all sort of map strengths of relations with your input words & patterns to some associated results that it just sort of statistically was reinforced with. These can yield information that is of real use for a human to further evaluate and act on, but I wouldn't trust such a system to directly act on those associations; they're wrong too often.
But there is no possibility of actual conceptual discourse with NNs as we know them, to correct them, to inform them, or to ask them to explain their results. This is a fundamental barrier to achieving human-style intelligence.
This is not to say that there aren't NN-based possibilities that might work, like tacking together multiple interacting NNs, each which have the possibility to specialize on concepts and influence other NNs. But too much has focused on single NNs and direct input-to-output "monkey-see, monkey-do" training. It's a manifestation of the Chinese room problem.