10 ms·

To Build Truly Intelligent Machines, Teach Them Cause and Effect (2018)

146 points by guybedo 7y ago

uoaei 7y ago
Causal inference is the next big leap in AI. Once the relatively(!) low-hanging fruit of pattern recognition are picked to exhaustion, and once we can get more comfortable with symbolic reasoning with respect to theorem proving / hypothesis testing / counterfactuals, "real" reasoning machines will arise.
- ivalm 7y ago
  This sounds like going back to AI of the 80s. IMHO, symbolic reasoning is unlikely to lead to progress.
  - uoaei 7y ago
    It's not an either-or question. Symbolic reasoning comes up in many places where regression techniques are simply inadequate. A synthesis of the two is likely to see progress.
- mrfusion 7y ago
  I’d like to hear more about this. How do you see this coming about?
  - hummel 7y ago
    I'm working extensively on it, but I can´t give so many details yet.
  - uoaei 7y ago
    Definitely a combination of current ML (basically fancy nonlinear regression to MLE targets) with symbolic reasoning. Either alone are insufficient. Symbolic reasoning is basically learning a lot of "if-then" statements and chaining them to make inference. Causal reasoning consists of defining conditional dependencies of current state on past state, then extrapolating based on the encoded assumptions. It requires some notion of object relation both in a literal sense as well as subtler relationships. Regression techniques are being ham-fisted to fit these roles but the popular ML of today is still just pattern recognition and cannot be called "reasoning" per se. I don't work directly in this space but I see it following closely the architecture of the human brain for a while before departing to more distilled forms of knowledge management structures.
    - viuphiet 7y ago
      Neural networks do exactly what you are describing as "symbolic reasoning". It seems to be a common thing recently to dismiss modern ML techniques as curve fitting, but these fundamental models are extremely powerful. Neural networks are capable of approximating any system to arbitrary precision.
      - uoaei 7y ago
        This is theoretically true, but it's like saying "computers can compute any function, given enough time and resources". There is a need to construct logically-deduced models which impose an inductive bias so that your regression methods are efficient. That's where reasoning comes in, and where automated reasoning methods should be useful.
sarosh 7y ago
While the article is a nice Q&A with Pearl about his new book, The Book of Why, there is a very detailed technical tutorial from 2014 at http://research.microsoft.com/apps/video/default.aspx?id=206977 http://research.microsoft.com/apps/video/default.aspx?id=206... that provides a very in depth explanation of causal calculus / coutnerfactuals / etc. and how these tools should be used
- pieterk 7y ago
  Slides are here btw: https://media.nips.cc/Conferences/2013/nips-dec2013-pearl-bareinboim-tutorial-full.pdf https://media.nips.cc/Conferences/2013/nips-dec2013-pearl-ba...
bitxbit 7y ago
I strongly believe there’s overemphasis on AI, artificial intelligence, vs augmented intelligence.
- nightski 7y ago
  Do you have an indication that they are all that much different? Meaning, would the techniques or strategies used to develop augmented intelligence be that much different than what is going on in AI?
mindcrime 7y ago
There are a number of "things" that we should "teach" machines to create ones that are "truly intelligent". Besides this kind of "cause / effect reasoning", one could argue that an intelligent machine needs some baseline levels of what you might call "intuitive metaphysics", and "intuitive epistemology". You could probably argue that the cause/effect stuff is subsumed by one of these at a certain level of abstraction, but I think it makes sense to treat them as separate. Related to the idea of "cause/effect" and possibly falling into the overall rubric of "intuitive metaphsyics" is some notion of the passage of time. That is, in human experience we link things as "causal" when they happen in a certain sequence, and within a certain degree of temporal proximity. Eg, "I touched the hot burner then instantaneously felt excruciating pain" is an experience that we learn from. "I walked through the door and four days later I felt pain in my knee" probably is not. Our machines probably also need baseline levels of some sort of intuitive versions of Temporal Logic and Modal Logic as well. https://en.wikipedia.org/wiki/Metaphysics https://en.wikipedia.org/wiki/Metaphysics https://en.wikipedia.org/wiki/Epistemology https://en.wikipedia.org/wiki/Epistemology https://en.wikipedia.org/wiki/Temporal_logic https://en.wikipedia.org/wiki/Temporal_logic https://en.wikipedia.org/wiki/Modal_logic https://en.wikipedia.org/wiki/Modal_logic
- Barrin92 7y ago
  I'd agree with that and I think winograd schema make this very obvious, take for example: (1) John took the water bottle out of the backpack so that it would be lighter. (2) John took the water bottle out of the backpack so that it would be handy What does it refer to in each sentence? It's very obvious that a machine that solves this must understand physics, have a rudimentary ontology about objects and human intuition and so on. I think it's straight-up sad how little progress there has been on these very fundamental problems which articulate what common sense and intelligent agents are about.
  - foota 7y ago
    Hah. I did some experimenting with language model tasks here. I re-phrased the sentence as "John took the water bottle out of the backpack in order to make lighter|less heavy|handy the" etc. The only way it would complete with something other than the water bottle was "John took the water bottle out of the backpack in order to lighten the " and it completed with 'load' The most amusing was a task to fill in the blank with "John took the water bottle out of the backpack so that [MASK] would be lighter" The model was 98% confident the blank should be 'it' facepalm Interestingly enough when I change the fill in the blank sentence to: "John took the water bottle out of the backpack so that the [MASK] would be lighter." The result was: "23.9% liquid 13.4% water 7.7% contents 6.5% weight" and the last two are pretty close. I ran these tests on https://demo.allennlp.org/ https://demo.allennlp.org/.
sgt101 7y ago
Interesting formulation - because I think children learn about cause and effect. any hooo....
tabtab 7y ago
I'm not sure that's necessary. Early humans didn't know why a lot of things happened, such as why rubbing sticks makes fire; they just learned to use them from trial and error. The physics of it were beyond them. I see it more as goal-oriented: "I want fire, how can I get it?". I suppose that's cause-and-effect in a loose sense, but one doesn't have to view everything as C&E to get similar results. It seems more powerful to think of it as relationships instead of just C&E because then you get a more general relationship processing engine out of it instead of a single-purpose thing. Make C&E a sub-set of relationship processing. If the rest doesn't work, then you still have a C&E engine from it by shutting off some features.
- mindcrime 7y ago
  I suppose that's cause-and-effect in a loose sense, but one doesn't have to view everything as C&E to get similar results. It seems more powerful to think of it as relationships instead of just C&E because then you get a more general relationship processing engine out of it instead of a single-purpose thing. Make C&E a sub-set of relationship processing. If the rest doesn't work, then you still have a C&E engine from it by shutting off some features. I may be wrong (heck, I'm probably wrong), but I can't help but feel that you're abstracting things out too much. Yes, a "cause / effect relationship" IS-A "relationship", but sometimes the distinctions actually matter. I'd argue that a "cause/effect relationship" (and the associated reasoning) is markedly different in at least one important sense, and that is that it includes time in two senses: direction, and duration. There's a difference between knowing that Thing A and Thing B are "somehow" related, and knowing that "Doing Thing A causes Thing B to happen shortly afterwards" or whatever. To may way of thinking, this is something like what Pearl is talking about here: The key, he argues, is to replace reasoning by association with causal reasoning. Instead of the mere ability to correlate fever and malaria, machines need the capacity to reason that malaria causes fever. That said, I do like your idea of trying to build the processing engine in such a way that you can turn features on and off, because I don't necessarily hold that "cause/effect" is the only kind of reasoning we need.
  - tabtab 7y ago
    I'm not against tagging a vector between two or more nodes as "probably causal" (cause related) in some sense (or maybe a "weighted causal"). It just shouldn't be the ONLY tag. Re quote: "The key, he argues, is to replace reasoning by association with causal reasoning. Instead of the mere ability to correlate fever and malaria, machines need the capacity to reason that malaria causes fever." Back to my original point, humans did just fine at "intelligence" before science came along. That's Step 3, we need step 2 first. Find correlations, and if it seems to be that part-1 happens before part-2, then the bot can infer something equivalent to a causal relationship. That may be the time element you are talking about. Perhaps it's a matter of interpreting what "causal" means. I see it as "finding relationships we can take advantage of to obtain our goals". Whether there is physics or chemistry behind a relationship is an unnecessary distraction (without other advances). Re: That said, I do like your idea of trying to build the processing engine in such a way that you can turn features on and off, because I don't necessarily hold that "cause/effect" is the only kind of reasoning we need. So we kind of agree. If it turns out I'm wrong and that we can make most of the bot work right with just causal relationships, then factor out the general purpose "graph processor" for efficiency and make it causal-centric.
hans1729 7y ago
related, a deepmind-paper i found fascinanting: https://arxiv.org/pdf/1901.08162v1.pdf https://arxiv.org/pdf/1901.08162v1.pdf
Animats 7y ago
In the 1980s, when everybody was trying to do AI with some flavor of predicate calculus, he extended that to probabilistic predicate calculus. That helped. But it didn't lead to common sense reasoning. The field is so stuck that few people are even trying. Working on common sense, defined as predicting what happens next from observations of the current state, is a classic AI problem on which little progress has been made. I used to remark that most of life is avoiding big mistakes in the next 30 seconds. If you can't do that, life will go very badly. Solving that problem is "common sense". It's not an abstraction. The other classic problem where the field is stuck is robotic manipulation in unstructured situations. McCarthy once thought, in the 1960s, that it was a summer project to do that. He wanted a robot to assemble a Heathkit TV set kit. No way. (The TV set kit was actually purchased, sat around for years, and finally somebody assembled it and put it in a student lounge at Stanford.) 50 years later, unstructured manipulation still works very badly. Watch the DARPA Humanoid Challenge or the DARPA Manipulation Challenge videos from a few years ago. Great PhD thesis topics for really good people. High-risk; you'll probably fail. Succeed, even partially, and you have a good career ahead.
- jfengel 7y ago
  I worked on AI-via-predicate-calculus, as a successor to Cyc, and I think the main thing I learned is that people are incredibly bad at predicate calculus. Even when we behave "logically", it's an after-the-fact rationalization for a conclusion we arrived at much faster with heuristics. When we think-about-thinking, or talk-about-thinking, we do so in the language of language, which quickly leads to logic. And that leads us to think that the logic is the thinking. But in fact it's a rare, specialized mode of thought. The primary mode of thought -- the one that keeps us from making big mistakes for a half-minute at a time -- is that irrational one that's very easy to fool if you put effort into it, but which actually gets it right for most of reality (which isn't, generally, trying to trick you).
  - Animats 7y ago
    When we think-about-thinking, or talk-about-thinking, we do so in the language of language, which quickly leads to logic. And that leads us to think that the logic is the thinking. But in fact it's a rare, specialized mode of thought. Yes. Language is not thinking. Language is I/O.
neaden 7y ago
It's funny that I knew who this would be by or interviewing just from the title. I like Judea Pearl and a lot of his ideas, but at the same time I think he overstates their importance and hypes them up more then he should.
jacobwilliamroy 7y ago
I can build an AI with common sense reasoning in about 9 months. The problem has already been solved. Why do we care so much about making computers more like people? Isn't that excessively cruel? Part of the utility of computing is that computers don't have needs for fulfillment, companionship, communion. We deploy them in awful conditions to do the most horrible, tedious time-waster jobs. Why do such minds need to be human?
- nineteen999 7y ago
  I wonder how it could even be considered "cruel"? Cruel to other living human beings, perhaps. To the machine or its simulation software? No. Any human-like AI is still a "fake" - any notion of emotion, pain, empathy etc. we attribute to them is only a simulation. It simply doesn't matter. It amazes and amuses me to think that people might actually give a damn what the machine is "feeling". I think people who truly believe this are out of touch with reality and frankly, with other human beings. The machine doesn't really care about us, it's a bunch of ones and zeroes no matter how you slice and dice it. Even after training them on cause and effect, they still don't care. I don't buy the "if it looks like a human, sounds like a human, it's human" argument at all.
  - jacobwilliamroy 7y ago
    Sure it is. Sure it is. Whatever helps you sleep at night.
  - JoeAltmaier 7y ago
    Cruel is also in the mind of the one doing the cruelty. People worry about being cruel to plant, to pets, to their cars. Its natural and normal, because we are empathetic beings. Not something to try to unlearn or avoid; its a big part of our humanity.
    - nineteen999 7y ago
      A plant or an animal, yes. A car though? Only reason to worry about being "cruel" to a car is that mistreating it will result in larger repair bills and a need to replace it earlier. Same thing with a computer. But each to their own I guess.
Pils 7y ago
I recently joined a team that does a lot of causal analysis, mostly marketing related, and was wondering what the best resources are to get more familiar with this subject (books, lectures, online courses etc.). I am picking up the author's other book, Causality: Models, Reasoning and Inference, but wondering what other sources people recommend.
- mindcrime 7y ago
  Maybe a book or two on Structural Equation Modeling? https://en.m.wikipedia.org/wiki/Structural_equation_modeling https://en.m.wikipedia.org/wiki/Structural_equation_modeling
crimsonalucard 7y ago
Most people don't even know how to run an experiment to verify causation. They chant the mantra: "correlation does not equal causation" then go back to correlating everything they see in the world.
spappletrap 7y ago
"Teach them cause and effect" ... yep, that's pretty much what everyone's been trying to do since the 60's. The problem is that nobody will touch the core issues of consciousness because it's inherently political. It requires confronting some of the biggest taboos in science: anthropomorphizing animals in biology, discussing consciousness seriously in physics, and looking at how economics and information interact with a skeptical eye toward the standard economic narrative.
narag 7y ago
Simulating a mind is not the same as simulating mind processes. I doubt that you can create a mind that's similar to a human mind without the relevant elements that are took for granted when we think of a human being: senses, perception, pain, pleasure, fear, volition... a body! and the real-time feedback loop that connects us to our environment and our peers. The same could be said about animals' minds. That's why it's still impossible to make even a mosquito brain. It's a question of texture. Making a decision for a human involves a complex cloud of subsystems working in unstable equilibrium, more of a boiling cauldron than an algorithmic checklist. When you're scared, you're not just thinking that somethind is dangerous and rather avoid it, you are feeling something very uncomfortable and you want to stop it. What if you want to advance in creating some kind of simpler mind now when you still haven't the means to build a complete organism? That's an interesting problem. Would immersing programs in a virtual world be useful? Or would it be better to make robots face the real world directly? I believe that you need, as a minimum, a system that integrates sight with hearing and touching sensors, and some kind of incentive system. After some results, maybe using machine learning, the emergent organization could be applied as a building block to more complex robots. Meanwhile, trying to teach machines some human capabilities will not lead to generalized IA, but to more of the same we have now, that it's very useful, just not quite qualifies for the label.
- 7373737373 7y ago
  >senses, perception, pain, pleasure, fear, volition... a body! Yes! Almost all neural networks have no self-model and thus no self-awareness because they cannot perceive themselves. They only see the inputs. They do not see the result of their actions. This makes developing a self-model impossible. They cannot develop an internal model of internal vs external causes. What their "boundary of influence" is. Differentiation between internal and external causes. They are trained and then used, immutable, unlearning after training. Even if they could perceive their outputs during training and/or evaluation, they cannot perceive themselves otherwise, making it practically impossible to deduce by themselves what they even are. They can't inspect themselves. The causal loop needs to be closed for all of this to happen.
  - 7y ago
averros 7y ago
The do-calculus is definitely a step in the right direction (it does help to disentangle confounders and such). The major (and I mean major) shortcoming of do-calculus is that it still provides no framework for useful induction. You need to somehow come up with causal structure first so you could express it. And normally you don't have it, you only have observations and some a priori theory (which may be wrong or incomplete) of how things relate to each other. The real question is how do you come up with such theory in the first place. (The secondary question is ok, you have a theory, does the data support it? And if it cannot be determined from existing data, what do I need to perturb to get the data?)
RedComet 7y ago
A “truly intelligent machine” is a contradiction of terms. They can not have intelligence like humans (AGI or whatever the current buzzword is). Humans are not solely material.
- IIAOPSW 7y ago
  Not with that attitude. Seriously. Extraordinary claims etc etc. If you want to claim humans are not solely material, you need to give some sort of evidence of a phenomena beyond the physical. You can't use intelligence per se as your evidence as then your argument is circular.
  - RedComet 7y ago
    It has been demonstrated for millennia. And I don’t know what your strange intelligence straw man has to do with anything. Even elementary metaphysics covers this.
    - IIAOPSW 7y ago
      This is just neo-geocenterism. >Of course the Earth is in the center of the solar system. We must occupy a privileged space in this universe. Its been demonstrated for millennia. Thinking is done with neurons. Neurons are subject to the same physical laws as the rest of the material world. Therefore thinking can be done by a machine (if nothing else, by a physics simulation of neurons). To refute this logic, you must show some thought is not being done by neurons or that neurons are not subject to physics.
      - RedComet 7y ago
        I don't see why you feel the need to keep setting up these straw men. It might make for nice rhetoric, but it is more than a little disingenuous. Not only was that geocentrism piece not an argument or claim that I've made, but I've never seen anyone make it in that manner either. But perhaps you know that and were intentionally misrepresenting their arguments. Back on topic, no, your concluding claim is not true. To reach your conclusion you would, at the minimum, need to assume that a machine can simulate arbitrary physical phenomenon, which is not a forgone conclusion. For instance, the "thinking done by neurons" you refer to made be reliant on some facet of the real numbers that is simply not computable. Perhaps at any level of approximation of it, what we discern as AGI may not manifest. Etc. But, finally, your premises are faulty. What most people really mean when they reference AGI or "truly intelligent" is not intelligence, but wisdom. Computers have been "more intelligent" than humans for a long time now if it simply means arithmetic and recalling trivia. Now, noting that, isn't it possible that such wisdom is dependent on dependent on the will, the soul, etc? So we arrive back at the true point of contention - you are a materialist. I claim that materialism has been handily refuted for thousands of years. Then you fell back on pretty much every freshman level fallacy in the book. On the other hand, I suspect some of your misrepresenting of "classical" philosophy was not intentional, and just the result of getting most of it second hand from the Kurzweil (AI) and Dawkins (geocentrism) type literature. It is wise not to be so dismissive of pre-"Enlightenment" thinking.
hooande 7y ago
@dang, hate to be that guy, but can we add "[2018]" to the title?
- AlexCoventry 7y ago
  This was kind of passé, even in 2018.
agumonkey 7y ago
I'd say pain. But that's only me.
smiljo 7y ago
Saw only recently that Judea Pearl was a guest on Sam Harris' podcast: https://samharris.org/podcasts/164-cause-effect/ https://samharris.org/podcasts/164-cause-effect/. The preamble is depressing, since the episode aired right after a mass shooting, but Pearl gives a brief overview of his thinking.
ratsimihah 7y ago
Isn't reinforcement learning essentially a representation of cause and effect?
- AlexCoventry 7y ago
  No, it's a representation of which actions lead to good outcomes given a set of input data. There is no explicit symbolic reasoning about causal factors or their outcomes involved in classic RL, and it's very unlikely that any such symbolic representation evolves implicitly under the hood. A neural net in an RL system is just a souped-up version of the tabular data used in the earliest RL systems.
  - viuphiet 7y ago
    The reinforcement learning framework is perfect for representing cause and effect. An agent could learn that in a state of no fire, taking an action of rubbing sticks together would transition into a state of having fire. This concept is formalized as learning the dynamics function.

146 points by guybedo 7y ago

Causal inference is the next big leap in AI. Once the relatively(!) low-hanging fruit of pattern recognition are picked to exhaustion, and once we can get more comfortable with symbolic reasoning with respect to theorem proving / hypothesis testing / counterfactuals, "real" reasoning machines will arise.

ivalm 7y ago
This sounds like going back to AI of the 80s. IMHO, symbolic reasoning is unlikely to lead to progress.
- uoaei 7y ago
  It's not an either-or question. Symbolic reasoning comes up in many places where regression techniques are simply inadequate. A synthesis of the two is likely to see progress.
mrfusion 7y ago
I’d like to hear more about this. How do you see this coming about?
- hummel 7y ago
  I'm working extensively on it, but I can´t give so many details yet.
- uoaei 7y ago
  Definitely a combination of current ML (basically fancy nonlinear regression to MLE targets) with symbolic reasoning. Either alone are insufficient. Symbolic reasoning is basically learning a lot of "if-then" statements and chaining them to make inference. Causal reasoning consists of defining conditional dependencies of current state on past state, then extrapolating based on the encoded assumptions. It requires some notion of object relation both in a literal sense as well as subtler relationships. Regression techniques are being ham-fisted to fit these roles but the popular ML of today is still just pattern recognition and cannot be called "reasoning" per se. I don't work directly in this space but I see it following closely the architecture of the human brain for a while before departing to more distilled forms of knowledge management structures.
  - viuphiet 7y ago
    Neural networks do exactly what you are describing as "symbolic reasoning". It seems to be a common thing recently to dismiss modern ML techniques as curve fitting, but these fundamental models are extremely powerful. Neural networks are capable of approximating any system to arbitrary precision.
    - uoaei 7y ago
      This is theoretically true, but it's like saying "computers can compute any function, given enough time and resources". There is a need to construct logically-deduced models which impose an inductive bias so that your regression methods are efficient. That's where reasoning comes in, and where automated reasoning methods should be useful.

sarosh 7y ago

While the article is a nice Q&A with Pearl about his new book, The Book of Why, there is a very detailed technical tutorial from 2014 at http://research.microsoft.com/apps/video/default.aspx?id=206977 http://research.microsoft.com/apps/video/default.aspx?id=206... that provides a very in depth explanation of causal calculus / coutnerfactuals / etc. and how these tools should be used

pieterk 7y ago
Slides are here btw: https://media.nips.cc/Conferences/2013/nips-dec2013-pearl-bareinboim-tutorial-full.pdf https://media.nips.cc/Conferences/2013/nips-dec2013-pearl-ba...

bitxbit 7y ago

I strongly believe there’s overemphasis on AI, artificial intelligence, vs augmented intelligence.

nightski 7y ago
Do you have an indication that they are all that much different? Meaning, would the techniques or strategies used to develop augmented intelligence be that much different than what is going on in AI?

mindcrime 7y ago

There are a number of "things" that we should "teach" machines to create ones that are "truly intelligent". Besides this kind of "cause / effect reasoning", one could argue that an intelligent machine needs some baseline levels of what you might call "intuitive metaphysics", and "intuitive epistemology". You could probably argue that the cause/effect stuff is subsumed by one of these at a certain level of abstraction, but I think it makes sense to treat them as separate. Related to the idea of "cause/effect" and possibly falling into the overall rubric of "intuitive metaphsyics" is some notion of the passage of time. That is, in human experience we link things as "causal" when they happen in a certain sequence, and within a certain degree of temporal proximity. Eg, "I touched the hot burner then instantaneously felt excruciating pain" is an experience that we learn from. "I walked through the door and four days later I felt pain in my knee" probably is not. Our machines probably also need baseline levels of some sort of intuitive versions of Temporal Logic and Modal Logic as well. https://en.wikipedia.org/wiki/Metaphysics https://en.wikipedia.org/wiki/Metaphysics https://en.wikipedia.org/wiki/Epistemology https://en.wikipedia.org/wiki/Epistemology https://en.wikipedia.org/wiki/Temporal_logic https://en.wikipedia.org/wiki/Temporal_logic https://en.wikipedia.org/wiki/Modal_logic https://en.wikipedia.org/wiki/Modal_logic

Barrin92 7y ago
I'd agree with that and I think winograd schema make this very obvious, take for example: (1) John took the water bottle out of the backpack so that it would be lighter. (2) John took the water bottle out of the backpack so that it would be handy What does it refer to in each sentence? It's very obvious that a machine that solves this must understand physics, have a rudimentary ontology about objects and human intuition and so on. I think it's straight-up sad how little progress there has been on these very fundamental problems which articulate what common sense and intelligent agents are about.
- foota 7y ago
  Hah. I did some experimenting with language model tasks here. I re-phrased the sentence as "John took the water bottle out of the backpack in order to make lighter|less heavy|handy the" etc. The only way it would complete with something other than the water bottle was "John took the water bottle out of the backpack in order to lighten the " and it completed with 'load' The most amusing was a task to fill in the blank with "John took the water bottle out of the backpack so that [MASK] would be lighter" The model was 98% confident the blank should be 'it' facepalm Interestingly enough when I change the fill in the blank sentence to: "John took the water bottle out of the backpack so that the [MASK] would be lighter." The result was: "23.9% liquid 13.4% water 7.7% contents 6.5% weight" and the last two are pretty close. I ran these tests on https://demo.allennlp.org/ https://demo.allennlp.org/.

sgt101 7y ago

Interesting formulation - because I think children learn about cause and effect. any hooo....

tabtab 7y ago

I'm not sure that's necessary. Early humans didn't know why a lot of things happened, such as why rubbing sticks makes fire; they just learned to use them from trial and error. The physics of it were beyond them. I see it more as goal-oriented: "I want fire, how can I get it?". I suppose that's cause-and-effect in a loose sense, but one doesn't have to view everything as C&E to get similar results. It seems more powerful to think of it as relationships instead of just C&E because then you get a more general relationship processing engine out of it instead of a single-purpose thing. Make C&E a sub-set of relationship processing. If the rest doesn't work, then you still have a C&E engine from it by shutting off some features.

mindcrime 7y ago
I suppose that's cause-and-effect in a loose sense, but one doesn't have to view everything as C&E to get similar results. It seems more powerful to think of it as relationships instead of just C&E because then you get a more general relationship processing engine out of it instead of a single-purpose thing. Make C&E a sub-set of relationship processing. If the rest doesn't work, then you still have a C&E engine from it by shutting off some features. I may be wrong (heck, I'm probably wrong), but I can't help but feel that you're abstracting things out too much. Yes, a "cause / effect relationship" IS-A "relationship", but sometimes the distinctions actually matter. I'd argue that a "cause/effect relationship" (and the associated reasoning) is markedly different in at least one important sense, and that is that it includes time in two senses: direction, and duration. There's a difference between knowing that Thing A and Thing B are "somehow" related, and knowing that "Doing Thing A causes Thing B to happen shortly afterwards" or whatever. To may way of thinking, this is something like what Pearl is talking about here: The key, he argues, is to replace reasoning by association with causal reasoning. Instead of the mere ability to correlate fever and malaria, machines need the capacity to reason that malaria causes fever. That said, I do like your idea of trying to build the processing engine in such a way that you can turn features on and off, because I don't necessarily hold that "cause/effect" is the only kind of reasoning we need.
- tabtab 7y ago
  I'm not against tagging a vector between two or more nodes as "probably causal" (cause related) in some sense (or maybe a "weighted causal"). It just shouldn't be the ONLY tag. Re quote: "The key, he argues, is to replace reasoning by association with causal reasoning. Instead of the mere ability to correlate fever and malaria, machines need the capacity to reason that malaria causes fever." Back to my original point, humans did just fine at "intelligence" before science came along. That's Step 3, we need step 2 first. Find correlations, and if it seems to be that part-1 happens before part-2, then the bot can infer something equivalent to a causal relationship. That may be the time element you are talking about. Perhaps it's a matter of interpreting what "causal" means. I see it as "finding relationships we can take advantage of to obtain our goals". Whether there is physics or chemistry behind a relationship is an unnecessary distraction (without other advances). Re: That said, I do like your idea of trying to build the processing engine in such a way that you can turn features on and off, because I don't necessarily hold that "cause/effect" is the only kind of reasoning we need. So we kind of agree. If it turns out I'm wrong and that we can make most of the bot work right with just causal relationships, then factor out the general purpose "graph processor" for efficiency and make it causal-centric.

hans1729 7y ago

related, a deepmind-paper i found fascinanting: https://arxiv.org/pdf/1901.08162v1.pdf https://arxiv.org/pdf/1901.08162v1.pdf

Animats 7y ago

In the 1980s, when everybody was trying to do AI with some flavor of predicate calculus, he extended that to probabilistic predicate calculus. That helped. But it didn't lead to common sense reasoning. The field is so stuck that few people are even trying. Working on common sense, defined as predicting what happens next from observations of the current state, is a classic AI problem on which little progress has been made. I used to remark that most of life is avoiding big mistakes in the next 30 seconds. If you can't do that, life will go very badly. Solving that problem is "common sense". It's not an abstraction. The other classic problem where the field is stuck is robotic manipulation in unstructured situations. McCarthy once thought, in the 1960s, that it was a summer project to do that. He wanted a robot to assemble a Heathkit TV set kit. No way. (The TV set kit was actually purchased, sat around for years, and finally somebody assembled it and put it in a student lounge at Stanford.) 50 years later, unstructured manipulation still works very badly. Watch the DARPA Humanoid Challenge or the DARPA Manipulation Challenge videos from a few years ago. Great PhD thesis topics for really good people. High-risk; you'll probably fail. Succeed, even partially, and you have a good career ahead.

jfengel 7y ago
I worked on AI-via-predicate-calculus, as a successor to Cyc, and I think the main thing I learned is that people are incredibly bad at predicate calculus. Even when we behave "logically", it's an after-the-fact rationalization for a conclusion we arrived at much faster with heuristics. When we think-about-thinking, or talk-about-thinking, we do so in the language of language, which quickly leads to logic. And that leads us to think that the logic is the thinking. But in fact it's a rare, specialized mode of thought. The primary mode of thought -- the one that keeps us from making big mistakes for a half-minute at a time -- is that irrational one that's very easy to fool if you put effort into it, but which actually gets it right for most of reality (which isn't, generally, trying to trick you).
- Animats 7y ago
  When we think-about-thinking, or talk-about-thinking, we do so in the language of language, which quickly leads to logic. And that leads us to think that the logic is the thinking. But in fact it's a rare, specialized mode of thought. Yes. Language is not thinking. Language is I/O.

neaden 7y ago

It's funny that I knew who this would be by or interviewing just from the title. I like Judea Pearl and a lot of his ideas, but at the same time I think he overstates their importance and hypes them up more then he should.

jacobwilliamroy 7y ago

I can build an AI with common sense reasoning in about 9 months. The problem has already been solved. Why do we care so much about making computers more like people? Isn't that excessively cruel? Part of the utility of computing is that computers don't have needs for fulfillment, companionship, communion. We deploy them in awful conditions to do the most horrible, tedious time-waster jobs. Why do such minds need to be human?

nineteen999 7y ago
I wonder how it could even be considered "cruel"? Cruel to other living human beings, perhaps. To the machine or its simulation software? No. Any human-like AI is still a "fake" - any notion of emotion, pain, empathy etc. we attribute to them is only a simulation. It simply doesn't matter. It amazes and amuses me to think that people might actually give a damn what the machine is "feeling". I think people who truly believe this are out of touch with reality and frankly, with other human beings. The machine doesn't really care about us, it's a bunch of ones and zeroes no matter how you slice and dice it. Even after training them on cause and effect, they still don't care. I don't buy the "if it looks like a human, sounds like a human, it's human" argument at all.
- jacobwilliamroy 7y ago
  Sure it is. Sure it is. Whatever helps you sleep at night.
- JoeAltmaier 7y ago
  Cruel is also in the mind of the one doing the cruelty. People worry about being cruel to plant, to pets, to their cars. Its natural and normal, because we are empathetic beings. Not something to try to unlearn or avoid; its a big part of our humanity.
  - nineteen999 7y ago
    A plant or an animal, yes. A car though? Only reason to worry about being "cruel" to a car is that mistreating it will result in larger repair bills and a need to replace it earlier. Same thing with a computer. But each to their own I guess.

Pils 7y ago

I recently joined a team that does a lot of causal analysis, mostly marketing related, and was wondering what the best resources are to get more familiar with this subject (books, lectures, online courses etc.). I am picking up the author's other book, Causality: Models, Reasoning and Inference, but wondering what other sources people recommend.

mindcrime 7y ago
Maybe a book or two on Structural Equation Modeling? https://en.m.wikipedia.org/wiki/Structural_equation_modeling https://en.m.wikipedia.org/wiki/Structural_equation_modeling

crimsonalucard 7y ago

Most people don't even know how to run an experiment to verify causation. They chant the mantra: "correlation does not equal causation" then go back to correlating everything they see in the world.

spappletrap 7y ago

"Teach them cause and effect" ... yep, that's pretty much what everyone's been trying to do since the 60's. The problem is that nobody will touch the core issues of consciousness because it's inherently political. It requires confronting some of the biggest taboos in science: anthropomorphizing animals in biology, discussing consciousness seriously in physics, and looking at how economics and information interact with a skeptical eye toward the standard economic narrative.

narag 7y ago

Simulating a mind is not the same as simulating mind processes. I doubt that you can create a mind that's similar to a human mind without the relevant elements that are took for granted when we think of a human being: senses, perception, pain, pleasure, fear, volition... a body! and the real-time feedback loop that connects us to our environment and our peers. The same could be said about animals' minds. That's why it's still impossible to make even a mosquito brain. It's a question of texture. Making a decision for a human involves a complex cloud of subsystems working in unstable equilibrium, more of a boiling cauldron than an algorithmic checklist. When you're scared, you're not just thinking that somethind is dangerous and rather avoid it, you are feeling something very uncomfortable and you want to stop it. What if you want to advance in creating some kind of simpler mind now when you still haven't the means to build a complete organism? That's an interesting problem. Would immersing programs in a virtual world be useful? Or would it be better to make robots face the real world directly? I believe that you need, as a minimum, a system that integrates sight with hearing and touching sensors, and some kind of incentive system. After some results, maybe using machine learning, the emergent organization could be applied as a building block to more complex robots. Meanwhile, trying to teach machines some human capabilities will not lead to generalized IA, but to more of the same we have now, that it's very useful, just not quite qualifies for the label.

7373737373 7y ago
>senses, perception, pain, pleasure, fear, volition... a body! Yes! Almost all neural networks have no self-model and thus no self-awareness because they cannot perceive themselves. They only see the inputs. They do not see the result of their actions. This makes developing a self-model impossible. They cannot develop an internal model of internal vs external causes. What their "boundary of influence" is. Differentiation between internal and external causes. They are trained and then used, immutable, unlearning after training. Even if they could perceive their outputs during training and/or evaluation, they cannot perceive themselves otherwise, making it practically impossible to deduce by themselves what they even are. They can't inspect themselves. The causal loop needs to be closed for all of this to happen.
- 7y ago

averros 7y ago

The do-calculus is definitely a step in the right direction (it does help to disentangle confounders and such). The major (and I mean major) shortcoming of do-calculus is that it still provides no framework for useful induction. You need to somehow come up with causal structure first so you could express it. And normally you don't have it, you only have observations and some a priori theory (which may be wrong or incomplete) of how things relate to each other. The real question is how do you come up with such theory in the first place. (The secondary question is ok, you have a theory, does the data support it? And if it cannot be determined from existing data, what do I need to perturb to get the data?)

RedComet 7y ago

A “truly intelligent machine” is a contradiction of terms. They can not have intelligence like humans (AGI or whatever the current buzzword is). Humans are not solely material.

IIAOPSW 7y ago
Not with that attitude. Seriously. Extraordinary claims etc etc. If you want to claim humans are not solely material, you need to give some sort of evidence of a phenomena beyond the physical. You can't use intelligence per se as your evidence as then your argument is circular.
- RedComet 7y ago
  It has been demonstrated for millennia. And I don’t know what your strange intelligence straw man has to do with anything. Even elementary metaphysics covers this.
  - IIAOPSW 7y ago
    This is just neo-geocenterism. >Of course the Earth is in the center of the solar system. We must occupy a privileged space in this universe. Its been demonstrated for millennia. Thinking is done with neurons. Neurons are subject to the same physical laws as the rest of the material world. Therefore thinking can be done by a machine (if nothing else, by a physics simulation of neurons). To refute this logic, you must show some thought is not being done by neurons or that neurons are not subject to physics.
    - RedComet 7y ago
      I don't see why you feel the need to keep setting up these straw men. It might make for nice rhetoric, but it is more than a little disingenuous. Not only was that geocentrism piece not an argument or claim that I've made, but I've never seen anyone make it in that manner either. But perhaps you know that and were intentionally misrepresenting their arguments. Back on topic, no, your concluding claim is not true. To reach your conclusion you would, at the minimum, need to assume that a machine can simulate arbitrary physical phenomenon, which is not a forgone conclusion. For instance, the "thinking done by neurons" you refer to made be reliant on some facet of the real numbers that is simply not computable. Perhaps at any level of approximation of it, what we discern as AGI may not manifest. Etc. But, finally, your premises are faulty. What most people really mean when they reference AGI or "truly intelligent" is not intelligence, but wisdom. Computers have been "more intelligent" than humans for a long time now if it simply means arithmetic and recalling trivia. Now, noting that, isn't it possible that such wisdom is dependent on dependent on the will, the soul, etc? So we arrive back at the true point of contention - you are a materialist. I claim that materialism has been handily refuted for thousands of years. Then you fell back on pretty much every freshman level fallacy in the book. On the other hand, I suspect some of your misrepresenting of "classical" philosophy was not intentional, and just the result of getting most of it second hand from the Kurzweil (AI) and Dawkins (geocentrism) type literature. It is wise not to be so dismissive of pre-"Enlightenment" thinking.

hooande 7y ago

@dang, hate to be that guy, but can we add "[2018]" to the title?

AlexCoventry 7y ago
This was kind of passé, even in 2018.

agumonkey 7y ago

I'd say pain. But that's only me.

smiljo 7y ago

Saw only recently that Judea Pearl was a guest on Sam Harris' podcast: https://samharris.org/podcasts/164-cause-effect/ https://samharris.org/podcasts/164-cause-effect/. The preamble is depressing, since the episode aired right after a mass shooting, but Pearl gives a brief overview of his thinking.

ratsimihah 7y ago

Isn't reinforcement learning essentially a representation of cause and effect?

AlexCoventry 7y ago
No, it's a representation of which actions lead to good outcomes given a set of input data. There is no explicit symbolic reasoning about causal factors or their outcomes involved in classic RL, and it's very unlikely that any such symbolic representation evolves implicitly under the hood. A neural net in an RL system is just a souped-up version of the tabular data used in the earliest RL systems.
- viuphiet 7y ago
  The reinforcement learning framework is perfect for representing cause and effect. An agent could learn that in a state of no fire, taking an action of rubbing sticks together would transition into a state of having fire. This concept is formalized as learning the dynamics function.