This is getting absurd. An interview with Dr. Rodney Brooks, one of the great minds working in CS and robotics, has to spend time rebutting uninformed claims and fear mongering about AI research. There is so much Brooks can teach us. I look forward to reading his book.
Over the past seven years most of my time has been spent in robotics research labs, and I really struggle to reconcile the state of research with the concerns of those like Elon Musk. I think a series of discussions between the major figures on each side of this would be really valuable.
And Brooks doesn't understand Musk. He's not saying current AI is a threat. He's talking about the very long term future. What AI will be like in 30 years, or even further.
Its inevitable we will eventually solve AI. And when that day comes it will be dangerous. How easy do you think it is to control a being thousands of times smarter than you? If it was invented today we would have no ability to control it. Our best AI control mechanisms are just pressing a button to reward or punish it for it's behavior. You can't imagine any way that would fail?
Our slightly larger brains made the difference between swinging in trees and walking on the moon. But we are only the very first intelligence to evolve. Its unlikely we are anywhere near the peak of what is possible.
And this will likely happen in our lifetimes. The median expected date estimated by AI researchers is in the 2040s. Sure they can't possibly predict it very well, but who else can? And there is something to the wisdom of crowds.
Because there's nothing magical about the human brain. Evolution created it through just dumb mutation and selection. Under a bunch of ridiculous constraints. Like it had to use less than 10 watts of power, and weigh only a handful of pounds, and it had to be made of meat, and could only be iteratively improved from whatever happened to work at guiding locomotion in fish, etc.
Our current transistor tech is already orders of magnitude smaller, faster, and more efficient than neurons. There's no reason to expect the software stats of the brain are much better.
Interestingly, about 18% of the researchers said "never", but these answers didn't influence the posted CIs. Any reasonable imputation (say, "never = 2350") would both stretch those intervals by a lot and still underestimate the expert consensus.
They did account for them in the median AFAIK. Which is the number I gave and the most appropriate metric.
I don't see any mention of confidence intervals mentioned in the survey, so what are you talking about? Means and standard deviation are given. But they are mostly useless as the distribution is very skewed. They would be infinite if the "never" people were included of course.
The problem isn't just emulating the IQ of a typical human brain - which is a low bar, for the most part - but creating a transcendentally smart general intelligence.
Considering that we still have many issues with basic computing - not least working out how to write code that works reliably without constant manual updates - the suggestion that we're going to start building self-improving crash-proof hyperbrains any time soon is wishful thinking.
Realistically, we don't have the first clue how to start solving that problem. As of 2017 "Throw deep learning at it" is more of a fashion statement than a practical solution.
> Because there's nothing magical about the human brain.
Perhaps not "magical", but the human brain exists in a very unique situation because it's the only one we're trying to understand with other instances of itself.
Agreed. Also there is something "magical". Nobody understands how our consciousness works. While people will tell you it "arises from the brain" nobody has the slightest clue about how that works. It's a really hard problem when you think about it because our subjective experience has no mass , no length, no height, and is in that way not even material. Our consciousness is very important to our function, and without it, all artificial intelligence remains artificial and has no ultimate "reality check" on whether it's decisions are good or bad for it.
There's nothing "magical" about the human brain, but it's an extremely large, massively parallel device that with a clock speed of about 100 Hz.
When I say "massively parallel" I mean that a human brain (forget rest of nervous system) has about 80 G neurons. Each of which has a fan out between 10^4 and 10^5.
Architecturally this is quite different from a transistorized device with a much faster clock speed but a paucity of interconnect. Now few people say that we need to duplicate the brain (Numenta excepted -- although I have been half of a team that made a new brain and it was fun). After all an automobile does not look like a cheetah nor a plane look like an eagle. But this should give you some idea of the complexity of the problem.
(At Leela we are building something that approaches a more "brain" like computation when compared to, say, deep Q learning, and even for us, the equivalent of 2 M neurons requires about 16 TB of RAM).
- People who don't understand AI are afraid of it; those who do know how fragile and limited it is.
- The call for "regulation" of technology that doesn't exist is too vague to be useful.
- It's the AI in self-driving cars which has the most potential to immediately kill/save thousands of people, and it's telling that it's not this technology that Elon seems to be calling to be regulated. Whether or not regulation is the right thing to do, any argument for/against regulation of self-driving cars could be applied just the same to a hypothetical super AI, but the former is tied to real, practical problems which exist today.
Brooks clearly knows what's up.
Also, to add my own commentary:
- The dystopian robot future we should all be afraid of is not the [paperclip maximizer](https://wiki.lesswrong.com/wiki/Paperclip_maximizer) Musk and friends wave their arms about, but marketing/business algorithms that have ripple effects at the scale of societies-- the Facebook, YouTube, and Google ranking algorithms are examples of this. We could shortly be in a place where large scale human behavior is shaped by algorithms with more data and insight about collective human behavior than any single human could have, and it will be used to optimize for money making instead of stability, fairness, or cultural values. Some society-shaping decisions/policies could even be made without any human awareness of the reasoning behind them. This is not less scary if they're being made by fragile/flaky algorithms.
You don't need AI to have a paperclip demon. You just need complex systems that impact people's lives and aren't well understood. Capitalism is arguably the example of the paperclip deamon, and that will not get better with the permeation of sophisticated, fragile, poorly tuned machine learning techniques.
"Fragile and limited" does not mean safe. It is a strange idea that AI would need to be any of robust, general or complete to be risky. If anything, the main take-away from today's AI is that many problems that previously looked hard turned out to be solvable at superhuman levels with such unsophisticated machinery. This realisation should be extrapolated.
The call for regulation also need not be specific to be useful, though I suspect it would help. The greatest hurdle with AI risk is getting acceptance that this is a problem that needs to be dealt with ahead-of-time. Even if Elon does nothing but improve awareness, he has been useful.
Self-driving cars are actually the least in need of extra regulation. They are already regulated, their effects are observable, there is market pressure for them to perform along lines beneficial to humanity, there is little to no incentive to extend their AI to anything more general, etc. My expectation is that the AI that is most dangerous is the AI developed behind closed doors for purely private interests on broad domains.
I personally dislike the paperclip maximizer analogy; although it serves as a meaningful explanation of the AI alignment problem, people take the literal meaning too seriously, and the absurdity of it discredits the actual risk.
> The greatest hurdle with AI risk is getting acceptance that this is a problem that needs to be dealt with ahead-of-time. Even if Elon does nothing but improve awareness, he has been useful
Or he's distracted people from real problems, spreading fear and misinformation along the way, while real experts are sitting around without anyone listening to them because the god king elon is talking.
To me, Elon is playing catch-up and thrown his hat in the Yudkowsky club although Bostrom and other more credentialled people were probably his vector into it. They were talking about this stuff 10-20 years ago. I haven't yet seen anything where Elon moves the ball forward conceptually, although he's a doer so he's not messing around with writing futurism documents and nitpicking details of rationality that almost no one is actually capable of implementing.
On the other hand, Brooks doesn't show any indication he knows what Musk is talking about and throws out a bad summary of his position. I got nothing from this article except a slightly lower respect for Brooks.
The real minds to watch IMO is the Hinton + Deep Mind crew, and I think Yudkowsky and the fearmongers are largely correct, or at least correct enough to be taken seriously. I don't think people following the meme 'real AI researchers know that AI is limited and fragile' are on the right track. So that's my bias.
I think he understands fine, but he likes getting his name out there as an AI alarmist more.
> Tell me, what behavior do you want to change, Elon? By the way, let’s talk about regulation on self-driving Teslas, because that’s a real issue.
hahaha
This interview should be required reading for everyone who wants to talk about AI.
> It’s why I’m right now writing a book on AI and robotics and the future — because people are getting too scared about the wrong things and not thinking enough about what the real implications will be.
It's kind of amazing that we're at a point where "Rodney Brooks understands more about AI than Elon Musk" is news, but here we are. The power of PR is incredible.
24 comments
[ 3.7 ms ] story [ 60.1 ms ] threadOver the past seven years most of my time has been spent in robotics research labs, and I really struggle to reconcile the state of research with the concerns of those like Elon Musk. I think a series of discussions between the major figures on each side of this would be really valuable.
Its inevitable we will eventually solve AI. And when that day comes it will be dangerous. How easy do you think it is to control a being thousands of times smarter than you? If it was invented today we would have no ability to control it. Our best AI control mechanisms are just pressing a button to reward or punish it for it's behavior. You can't imagine any way that would fail?
Our slightly larger brains made the difference between swinging in trees and walking on the moon. But we are only the very first intelligence to evolve. Its unlikely we are anywhere near the peak of what is possible.
And this will likely happen in our lifetimes. The median expected date estimated by AI researchers is in the 2040s. Sure they can't possibly predict it very well, but who else can? And there is something to the wisdom of crowds.
Do you have references for this 2040 date? I'd love to see who's making this prediction.
Our current transistor tech is already orders of magnitude smaller, faster, and more efficient than neurons. There's no reason to expect the software stats of the brain are much better.
Here's one survey: http://www.nickbostrom.com/papers/survey.pdf
I don't see any mention of confidence intervals mentioned in the survey, so what are you talking about? Means and standard deviation are given. But they are mostly useless as the distribution is very skewed. They would be infinite if the "never" people were included of course.
Considering that we still have many issues with basic computing - not least working out how to write code that works reliably without constant manual updates - the suggestion that we're going to start building self-improving crash-proof hyperbrains any time soon is wishful thinking.
Realistically, we don't have the first clue how to start solving that problem. As of 2017 "Throw deep learning at it" is more of a fashion statement than a practical solution.
Perhaps not "magical", but the human brain exists in a very unique situation because it's the only one we're trying to understand with other instances of itself.
When I say "massively parallel" I mean that a human brain (forget rest of nervous system) has about 80 G neurons. Each of which has a fan out between 10^4 and 10^5.
Architecturally this is quite different from a transistorized device with a much faster clock speed but a paucity of interconnect. Now few people say that we need to duplicate the brain (Numenta excepted -- although I have been half of a team that made a new brain and it was fun). After all an automobile does not look like a cheetah nor a plane look like an eagle. But this should give you some idea of the complexity of the problem.
(At Leela we are building something that approaches a more "brain" like computation when compared to, say, deep Q learning, and even for us, the equivalent of 2 M neurons requires about 16 TB of RAM).
- People who don't understand AI are afraid of it; those who do know how fragile and limited it is.
- The call for "regulation" of technology that doesn't exist is too vague to be useful.
- It's the AI in self-driving cars which has the most potential to immediately kill/save thousands of people, and it's telling that it's not this technology that Elon seems to be calling to be regulated. Whether or not regulation is the right thing to do, any argument for/against regulation of self-driving cars could be applied just the same to a hypothetical super AI, but the former is tied to real, practical problems which exist today.
Brooks clearly knows what's up.
Also, to add my own commentary:
- The dystopian robot future we should all be afraid of is not the [paperclip maximizer](https://wiki.lesswrong.com/wiki/Paperclip_maximizer) Musk and friends wave their arms about, but marketing/business algorithms that have ripple effects at the scale of societies-- the Facebook, YouTube, and Google ranking algorithms are examples of this. We could shortly be in a place where large scale human behavior is shaped by algorithms with more data and insight about collective human behavior than any single human could have, and it will be used to optimize for money making instead of stability, fairness, or cultural values. Some society-shaping decisions/policies could even be made without any human awareness of the reasoning behind them. This is not less scary if they're being made by fragile/flaky algorithms.
AI isn't terrifying.
AI built with our current values is an horrific prospect.
The call for regulation also need not be specific to be useful, though I suspect it would help. The greatest hurdle with AI risk is getting acceptance that this is a problem that needs to be dealt with ahead-of-time. Even if Elon does nothing but improve awareness, he has been useful.
Self-driving cars are actually the least in need of extra regulation. They are already regulated, their effects are observable, there is market pressure for them to perform along lines beneficial to humanity, there is little to no incentive to extend their AI to anything more general, etc. My expectation is that the AI that is most dangerous is the AI developed behind closed doors for purely private interests on broad domains.
I personally dislike the paperclip maximizer analogy; although it serves as a meaningful explanation of the AI alignment problem, people take the literal meaning too seriously, and the absurdity of it discredits the actual risk.
Or he's distracted people from real problems, spreading fear and misinformation along the way, while real experts are sitting around without anyone listening to them because the god king elon is talking.
On the other hand, Brooks doesn't show any indication he knows what Musk is talking about and throws out a bad summary of his position. I got nothing from this article except a slightly lower respect for Brooks.
The real minds to watch IMO is the Hinton + Deep Mind crew, and I think Yudkowsky and the fearmongers are largely correct, or at least correct enough to be taken seriously. I don't think people following the meme 'real AI researchers know that AI is limited and fragile' are on the right track. So that's my bias.
Not quite true: "Tell me, what behavior do you want to change, Elon?"
> Tell me, what behavior do you want to change, Elon? By the way, let’s talk about regulation on self-driving Teslas, because that’s a real issue.
hahaha
This interview should be required reading for everyone who wants to talk about AI.
> It’s why I’m right now writing a book on AI and robotics and the future — because people are getting too scared about the wrong things and not thinking enough about what the real implications will be.
I want that book.