Hacker News Reader

Speaking as an AI researcher, almost nothing is "AI" research. In practice i feel most current AI research falls into two categories:

* Fuzzy problems -- image, sound and free text recognition. Where there is no real "true answer".

* Problems too hard to solve in a reasonable time without heuristics -- SAT, scheduling, etc. In practice NP-hard problems and further up the complexity heirachy -- AlphaGo goes here.

Once we know how to do something reliably, it stops being AI and just becomes "an algorithm" :)

Agree. Feng-hsiung Hsu, the chief designer of Deep Blue, the chess machine that beat Kasparov, once gave a talk at my university. I didn't go to the talk, but in the abstract, he stated clearly that he didn't think Deep Blue is AI -- it's just a search algorithm.

I don't understand why so many commenters here are so furious about the author's claim that AlphaGo is not AI. What qualifies as AI is really up to definition, and there doesn't seem to be any widely accepted one among AI researchers. The author of the IEEE article doesn't think AlphaGo matches his definition of AI. Other AI researchers may think otherwise. But that doesn't mean the author is trying to dismiss the achievement of the AlphaGo team.

>I don't understand why so many commenters here are so furious about the author's claim that AlphaGo is not AI.

I don't think they really are. I think they are annoyed because nobody (who knows the difference) ever claimed that AlphaGo was strong/true/general AI) - and the article feels like it's swatting at strawmen.

> Once we know how to do something reliably, it stops being AI and just becomes "an algorithm" :)

This.

AI is a moving definition and always seems to be "What we can't do right now."

is there any research into making a "generic" AI that can solve any problem without the researcher first having to know what that problem is? i.e., human style learning.

Certainly some very clever people are trying it, and have been since the 60s at least, but as far as i am aware the progress is limited (but it is not my area of expertise).

I think the main problem is you end up needing a language to describe the problem, and that ends up limiting the problems that can be solved, or you have to explain so carefully what the problem is it feels like cheating.

As a researcher in multiagent systems, I know this problem very well. Which is, I believe, exactly the point the article makes. But it goes on to say that this might be possible to overcome by embodiment and developmental psychology. Not new arguments for sure, but valid ones, and brought at the right time (a year ago) of a new AI hype.

Machine learning algorithms can solve many problems without knowing what they really are, only given examples (sometimes even without labels when unsupervised learning applies).

We are still quite far from anything we could call "human-style learning" though, but definitely getting there (just look at all the recent publications with reinforcement learning and elaborate ways to use memory in neural nets).

I think the parent is referring less to a system that can construct a single model without an explicit schema, and more to the thing that'd be a few steps after that—the ability to:

1. dynamically notice world-features ("instrumental goal features") that seem to correlate with terminal reward signals;

2. build+train entirely new contextual sub-models in response, that "notice" features relevant to activating the instrumental-goal feature;

3. shape goal-planning in terms of exploiting sub-model features to activate instrumental goals, rather than attempting to achieve terminal preferences directly. (And maybe also in terms of discovering sense-data that is "surprising" to the N most-useful sub-models.)

In other words, the AI should be able to interact with reward-stimuli at least as well as Pavlov's dog.

Right now, ML research does include the concept of "general game-playing" agents—but AFAIK, these agents are only expected to ever play one game per instance of the agent, with the generality being in how the same algorithm can become good at different games when "born into" different environments.

Humans (most animals, really) can become good at far more than a single game, because biological minds seem to build contextual models, that communicate with—but don't interfere with—the functioning of the terminal-preference-trained model.

So: is anyone trying to build an AI that can 1. learn that treats are tasty, and then 2. learn to play an unlimited number of games for treats, at least as well as a not-especially-smart dog?

Not entirely sure that is a fair description of human-style learning. Our overall problem to solve is 'survive and reproduce'. Anything else can be seen as just a sub-problem of that. Humans are taught by other humans how to solve problems since the day they are born. Our DNA passes on a millions of generations worth of learning from our ancestors about how to solve problems.

It is a fair description. That being: able to enumerate a large number of arbitrary goals and define a large number of basic pattern classifiers/feature extractors.

When people give credit to the human designers for AlphaGo's wins, that it is really a win for humanity, I disagree. The wins are alphago's even if the design is of human ingenuity.

When You say that the outputs of human ingenuity should be credited to Evolution, I similarly disagree. You might as well credit evolution for AlphaGo's win. While it is true that Evolution invented the first AGI (and in some though not all ways, a superior intelligence to it), it still makes sense to separate the products of human learning from whatever structural priors DNA passed along. I'll also point out that compared to most animals, humans actually have weaker priors and spend a lot of their early days learning to learn.

Tangential quibble: For anything that could be called "human-style" learning, which presumably requires abstract intelligence and cultural transmission, we're probably looking at something that has existed only during the period of human behavioral modernity - commonly taken to date from roughly 40-50,000 years ago. Assuming a generation is ~20 years (and there are some arguments that the average generation might have been closer to 25 years), that's only 2500 generations.

I think it's fascinating that we have developed from un-self-reflective animals, to abstract thinkers on the verge of creating wholly new abstract-thinking entities from scratch, in only two and half thousand steps. Especially given that the majority of the technical knowledge necessary was developed only in the last 500 years, or 25 generations.

Humans need just a bit more supervision than that.

> It is by the general and widely used definition that AlphaGo is actually an AI.

Only for a little while. AI is how we describe new technology that does something we think of as requiring human intelligence. AI, like the word "technology" itself, is a term we use during the lag between creating something and learning to take it for granted.

I don't foresee an end to technological development, but I do think eventually there will be no AI, because our ability to intuitively appreciate intelligence doesn't much surpass our own. Beyond some point we will stop perceiving new technological capabilities as higher levels of intelligence. There will be people, computers, appliances, services not tied to individual machines, and genetically engineered rabbit tutors for our children, and nobody will be surprised to encounter intelligence anywhere, least of all in a computer.

When some people say "AI" what they really mean is "Strong AI"

I think people think of it more along the lines of an artificial consciousness.

Which since consciousness is likely an emerging property of our own biological neural net and training data/weights, would likely have similar issues that humans have when exposed to bad data while growing up.

> our own biological neural net and training data/weights

Well that shoots down my "minds are made of mechanical internal drives/pistons/levers" theory.

I'm some people and what I mean by AI is artificial intelligence. If needed I can be more precise with Weak / Strong / AGI or ASI.

So now you speak for everyone, and define their language usage independent of their intended connotations? Quite presumptuous.

You are some other people, then.

What's a "weak AI"? Does any algorithm that is able to adapt its behavior by learning qualify as "weak AI"? If so, why call it AI at all?

"the capability of a machine to imitate intelligent human behavior"

That's what I would define as weak AI. So yes, even algorithms which do not adapt their behaviors by learning, but can imitate intelligent human behavior to some extent would be weak AI, such as AI in video games.

Now strong AI would be: "the capability of a machine to originate intelligent human behavior"

This means the intelligent behavior would originate from the machine itself. Some Machine Learning algorithms exhibit strong AI, in that they find ways to even outperform a human, by themselves, through the learning process. Sometimes, we can't even understand why it chose to do what it did, but it appeared to be even more intelligent than what most human would have chosen to do themselves.

Now Artificial General Intelligence, or sometimes called strong AI also, but I think there needs to be a distinction as I explain before. This would be: "the capability of a machine to originate any intelligent human behavior"

The reason a lot of people are talking about true AI, AGI, strong AI, etc. today, is because Machine Learning is seen as the be all, end all of AI. Yet practitioners and scientists know that it is not, at least in its current form. Currently, it can solve simple decision tasks, probably the kind of decisions a person makes in under a second or two, like driving. But it does not solve them generally, so a person must train different models for each tasks.

At the same time, having a way to automate decision making of a single simple task is huge in terms of potential. There's so many things we can automate through this, and if we combine them, you can create even more complex autonomous machines. That's why people are excited. Though I understand that it is not as awesome as having found general AI, or AI which can make complex decisions. The dream of AI continues, but the practicality of it are closer then ever.

> So yes, even algorithms which do not adapt their behaviors by learning, but can imitate intelligent human behavior to some extent would be weak AI

Replacing various human thought capabilities has been the goal of computers and algorithms since Turing and von Neumann.

> in that they find ways to even outperform a human, by themselves, through the learning process

Computers have been outperforming humans in many tasks since the moment of their creation. Some animals learn some things better than humans. Are they more intelligent?

> Machine Learning is seen as the be all, end all of AI.

I think it is seen as that mostly by non-experts, and generally by those who just want to believe we're close.

> At the same time, having a way to automate decision making of a single simple task is huge in terms of potential. There's so many things we can automate through this, and if we combine them, you can create even more complex autonomous machines.

Sure, I just don't see what any of this has to do with intelligence. Bacteria are autonomous, and so is the weather. The universe, too, for that matter. None of those is particularly intelligent.

Yeah, exactly. We call it AI because we need a name for this kind of adaptive algorithm, even if it isn't anything close to strong AI

There are JIT compilers, and even sorting algorithms that are adaptive. Are they AI too? What's wrong with the name "statistical learning" (as that's the flavor du jour of so called AI anyway)?

I think the word "adaptive" is meant to mean "general" learning that can be applied to more behavior.

AlphaGo was trained for blackgammon, could it be given chess and learn it just as well? The answer is no.

If I had to personally define AI, I would give a machine input to limbs of robot, and output of cameras and microphones (eyes and ears). And I would see if given sufficient years, can it learn language and conversation and ability to use robotic body for sports.

No, that is totally ignoring the point of the article due to it's semantics.

The point the article makes is that a general intelligence (or however you wanna call something that has intelligent function like a human) is not attained by AlphaGo. It might help but it is only part of the puzzle. There is work needed in other directions.