Slashdot Mirror
Wired Founding Editor Now Challenges 'The Myth of A Superhuman AI' (backchannel.com)
Wired's founding executive editor Kevin Kelly wrote a 5,000-word takedown on "the myth of a superhuman AI," challenging dire warnings from Bill Gates, Stephen Hawking, and Elon Musk about the potential extinction of humanity at the hands of a superintelligent constructs. Slashdot reader mirandakatz calls it an "impeccably argued debunking of this pervasive myth." Kelly writes:
Buried in this scenario of a takeover of superhuman artificial intelligence are five assumptions which, when examined closely, are not based on any evidence...
1.) Artificial intelligence is already getting smarter than us, at an exponential rate.
2.) We'll make AIs into a general purpose intelligence, like our own.
3.) We can make human intelligence in silicon.
4.) Intelligence can be expanded without limit.
5.) Once we have exploding superintelligence it can solve most of our problems...
If the expectation of a superhuman AI takeover is built on five key assumptions that have no basis in evidence, then this idea is more akin to a religious belief -- a myth
Kelly proposes "five heresies" which he says have more evidence to support them -- including the prediction that emulating human intelligence "will be constrained by cost" -- and he likens artificial intelligence to the physical powers of machines. "[W]hile all machines as a class can beat the physical achievements of an individual human...there is no one machine that can beat an average human in everything he or she does."
Kelly proposes "five heresies" which he says have more evidence to support them -- including the prediction that emulating human intelligence "will be constrained by cost" -- and he likens artificial intelligence to the physical powers of machines. "[W]hile all machines as a class can beat the physical achievements of an individual human...there is no one machine that can beat an average human in everything he or she does."
Anything is possible in 10-20 years, just give me all your money!
Because intelligence as a single-dimensioned parameter is a myth.
We already of have software with super-human information processing capabilities; and we're constantly adding more kinds of software that outperforms humans in specific tasks. Ultimately we'll have AIs that are as versatile has humans too. But "just as versatile" doesn't mean "good at the same things".
So it's probably true that software is getting smarter at exponential rates (and humans aren't getting smarter as far as I can see), but only in certain ways.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
The first AIs will be purpose built like today's supercomputers. They will make weather predictions, analyse financial trends, or study languages. Actually being intelligent isn't really necessary for interacting with humans, they only need to fake it well enough to fool us. The shift in society comes when those purpose-built AIs are efficiently linked along with the ability to interact with us. This is when it stops faking intelligence and actually becomes intelligent.
Given that our knowledge of the computational complexity of a single neuron is growing steadily, I think it's safe to say your FPGA cell estimate for a neuron was significantly too low.
For example, scientists now know that one single neuron (of certain types) is an entire neural network all by itself. Dendrites with multiple localized spikes communicating with each other and with other cells. Ultimately performing non-linear computation prior to forwarding any signal to cell body.
there is no one machine that can beat an average human in everything he or she does
Neither can most humans. There is no such thing as an average human. Every individual human specializes, and increasingly so as they get older (or they do not improve). It is a pervasive strawman to require AIs to "beat" an average human when the same quality isn't used to judge humans.
Those who do not learn from commit history are doomed to regress it.
... I'm glad I did not RTFA.
> 1.) Artificial intelligence is already getting smarter than us, at an exponential rate
Nobody who knows anything says that. We don't have real AI at all yet, just expert systems and a few interesting decision algorithms.
> 2.) We'll make AIs into a general purpose intelligence, like our own.
Of course we will. (Why would anyone make a phone that is also a web browser, a camera, an appointment tracker, a video game machine, a music player, a movie player, a flashlight, a compass, a map, a light level sensor, and a motion sensor?)
If you've figured out AI, you go general as soon as you can, because you get everything in one box.
>3.) We can make human intelligence in silicon
Meat is not special. In fact, we have a lot more reason to believe we'll be able to build an intelligence in silicon that is more efficient than evolution built with meat that to believe it's impossible because [insert magical thinking].
> 4.) Intelligence can be expanded without limit.
Lots of singularity nuts may think this, but again, anyone who knows anything about the universe will understand there must be a finite limit. We don't have any reason to believe humans are anywhere near it - and we could at least expect to make an AI as smart as the smartest human ever, and then take out the unnecessary bits that slowed that person down. Then up the clock rate.
> 5.) Once we have exploding superintelligence it can solve most of our problems
Most of the problems that can be solved with thought and not action and where cooperation with implementing the solution can be reasonably expected.
In short, Wired's founding executive editor Kevin Kelly is (at least in this instance) a buffoon speaking of things he does not understand sufficiently well to be speaking of them from a public platform.
The first three assumptions in this article have already been met sufficiently well enough to debunk the Wired article. AlphaGo has displayed superhuman intelligence in the first three areas of assumptions. 1) AlphaGo exploded on the scene by beating world class Go players much faster and much earlier than expected. Exponentially is a loaded word. e^0.0000001 is an exponential growth rate. So let's not quibble about how exponential the growth rate is.
2) AlphaGo is a general purpose learning tool. Just listen to the lectures and articles penned by the DeepMind team.
3) Alphago has displayed human-like intelligence, as claimed by the Go professionals it has played. They have said that AlphaGo plays like a human player.
4) If you take the fourth assumption literally, AI's intelligence is going to expand infinitely. Talking about infinity in human terms is unreasonable. Yes, AI's intelligence will expand.
5) The fifth assumption can be argued many ways. Some problems are not solvable due to their paradoxical nature. Other problems are subjective and are uniquely unsolvable by some individuals, but not by all individuals. It is a matter of time before a general purpose AI program will solve subjective emotional problems. Whether all human beings accept the solutions is subjective and open to speculation.
The human population is composed of experts, with divisions of labor. It is not unreasonable for AI programs to have areas of expertise.
Because then you wouldn't have been saying things like:
If you've figured out AI, you go general as soon as you can, because you get everything in one box.
...when Kelly dismisses that the concept of general-purpose AI because we look at intelligence through an anthropocentric lens. "General purpose" actually isn't.
AlphaGo is not a "general purpose" learning mechanism. It won't ever write sonnets meaningful to humans, or be able to to dance, or even employ symbolic differentiation.
It is a really nice toolset, and it is able to solve a task which is difficult for humans, but so does Google or your high-school calculator when you calculate sin(1.2).
It won't ever go beyond the computational underpinnings of playing Go-like games: evaluating game positions and calculating game trees. It won't ever say 'forget it, I'd rather be drinking beer with my buddies', which is an intelligent thing to do for most of us with respect to playing Go.
There's nothing human-like about AlphaGo, except that it solves a problem relevant to humans; the calculator example comes in mind.
I'd be thrilled to know what kind of specific major human problems you'd consider AI-approachable, because I currently only see a bunch of more or less advanced mechanisms that are fine-tuned to solve very specific computationally well-defined problems, and most human problems are not computationally well defined.
Your Points... #1. Fuel efficiency - The Lupo 3l, a Real world 78mpg, vs the model Ts reported 13-21... Yes I would say we are doing much better. Also considering that we have even made a plane that can fly around the world on NO fossil fuels of any kind. #2 Cheap Space Travel - Nobody said taking the entire environment of the Earth with you in simulated fashion was going to be cheap, or easy... That certainly does NOT mean we have made no discovery or achievement in space and exploration. Just the opposite. We are exploring FAR more of the observable universe than we EVER have, and the new wide field telescope is soon to completely change the way we look at the stars in the night sky and observable universe at large. #3 CPU Speed Plateau - Mostly correct, however, parallel computing and quantum computing are already changing the game. Making engineers think and program in radically new ways. The tides of change do ebb and flow. That does NOT mean they have some how halted. The laws of physics used to be something we could ONLY theorize, as we believed there was no real tangible way to TEST those theories. The LHC and CERN have shown us that this is not so. Same goes for the Photon and Graviton. Major Accomplishments in the modern era that change our understanding of physics on a "Daily" basis... So, yes and no... #4. Transportation speeds in the past 50-60 years in Both Air and Land Speed have BOTH had their bar raised MUCH higher than what was possible 50-60 years ago... By nearly 1000MPH in the Air since 1957, and around 360MPH on the Land in roughly the same timeframe.. So #4 is just plain False :-) New technologies do not "STOP" improving because a limit of physics has been hit... We simply start thinking in 3 dimensions or in radically new ways that the earth has never seen. Its all in the history books my friend. The Limitations of Physics are only limitations, because we do not yet fully understand the forces that created this Universe. But that too, is RAPIDLY changing. The Fields of Physics and Cosmology are discovering new tangible real world methods to verify the theory and turn it into facts we can work with. These perceived limitations are the result of a closed mind, not rooted in science. There are no limitations, just things we do not yet understand. Your knowledge of history is also lacking... Maybe Do your homework before trolling next time. K Anonymous? :-P Sources: http://www.motortrend.com/news... - - http://www.solarimpulse.com/ad... - http://www.gutenberg.cc/articl... - http://www.landspeedrecord.org...
I'm not sure whether it is discomfort at the idea of having a computer call them silly, a deep belief that humanity is somehow special in a special way (carefully defined in undefinable terms) or just a deep and enduring lack of imagination. Between AlphaGo beating Lee Sodel, the cancer treatments being proposed by Watson and the rise of driver-less cars we are seeing many supposedly impossible roles being taken over by software.
The five assumptions noted basically are basically denial ... reinforced with more denial. The evidence in a number of areas is in. Computers and software routinely appear in locations doing things predicted to be impossible. Computing capability keeps exceeding predictions.
Arguably the one valid assumption made is that intelligence is computable. If it is, the Church-Turing thesis gives the useful theoretical result that anything computable can be run on a UTM. It seems likely that what will end up happening is that the deniers keep arguing the point on what 'intelligence' is even after the AI they deny being possible has become bored with the discussion and moved on to more interesting pastimes.
If people like Bill Gates and Elon Musk are unrealistic in one direction, this person seems unrealistic in the other direction. He's basically betting against technological progress. And that's usually a losing bet, at least over long enough time periods.
1.) Artificial intelligence is already getting smarter than us, at an exponential rate.
Computers are already better than us at many tasks. That's been true for ages. And they're continuing to improve while we aren't. The set of tasks that computers are better at is constantly growing. I don't know of any fundamental limits to prevent them from eventually becoming better than us at the remaining tasks too. So it seems pretty likely they eventually will.
2.) We'll make AIs into a general purpose intelligence, like our own.
It's hard to even define what a "general purpose intelligence" means. But anything a human brain can do, computers will probably eventually be able to do it too.
3.) We can make human intelligence in silicon.
We can certainly make intelligence in silicon. We've already done it. Whether you consider it to be "human intelligence" or "inhuman intelligence" is kind of beside the point. If a computer can do something, whether it does it in the same way a human does is just an implementation detail.
4.) Intelligence can be expanded without limit.
I don't know of anyone who's claiming that. Where does he get this from? Anyway, the claim isn't that computers will advance without limit, only that they'll surpass humans.
5.) Once we have exploding superintelligence it can solve most of our problems...
Um, no. That's not at all what they're claiming. We certainly hope that it will solve many problems, but Gates, Musk, et al. are warning it could also create huge problems.
emulating human intelligence "will be constrained by cost"
Computers are cheap, and getting cheaper all the time. Humans are expensive and staying expensive. That's why automation has become such a big deal. Here again he seems to be betting against technological progress.
"I'm too busy to research this and form an educated opinion, but I do have time to tell everyone my uninformed opinion."
Unfortunately Sam Harris is bad at math. He claims "It's crucial to realize that the rate of progress doesn't matter, because any progress is enough to get us into the end zone. We don't need Moore's law to continue. We don't need exponential progress. We just need to keep going.". It seems he has never seen a monotonically increasing, yet asymptotically bounded function. However, that is exactly the kind of progress we are seeing in older technologies, e.g.: Airplanes stay at almost exactly same speed (because going past the sound barrier would use lots of energy) and get slightly more efficient each year, but will never get to the point where they can operate almost without any fuel or other large energy source, simply because the laws of physics don't allow that kind of progress.
But even if the possible progress is not bounded, it is still not guaranteed that we will get there. It can still take so long, that it never going to happen before human civilization is completely destroyed by some disaster. Or it could simply be stopped by economics as further improvements can easily get so expensive or tiny, that the likely benefits from pushing the research further can not offset the cost.
Harris also seems to think that general AI is ineviatable, because we want to make progress towards things such as things such as cureing cancer or Alzheimer. But it is not clear that such an achievement actually requires general superhuman intelligence. It likely requires superhuman intelligence, e.g.: the computers that simulate protein folding way better than any human could ever do, but not necessary general intelligence. Specialized artificial intelligence seems to be much easier to achieve and is at the same time likely almost as good as general intelligence for topics such as those. You don't need to develop an artificial general intelligence to cure cancer, if you already developed a specialized artificial intelligence that is able to find a cure.
Imagine what could happen when a huge neural net is applied.
The problem with huge neural nets is training them. The more possiblities a network has, the harder it becomes to train it. Large parts of the progress in the last few years were made by finding clever constraints on the network in order to make them easier to train.
Jan
Brains are not magic! The existence of human intelligence proves that intelligence is possible, everything else is just details.
Details can be damn hard to figure out and it is not so unlikely that evolution already found something that damn close to optimum if you consider factors such as energy to build and operate. It's tradeoffs are likely already getting tuned to changed environment, where high intelligence helps a lot and starvation isn't as big of an issue as it has been for millions for years. Potentially genetic engineering can make these changes quicker.
No, there will not be basic income or anything decent like that. There will be mass incarceration as people turn to crime to survive.
Mass incarceration is expensive and inefficient. It is likely much cheaper to pay for an basic income or a similar welfare system.
Jan
So, are you saying that reinforcement learning is intrinsically limited, or that AlphaGo is limited to the domain of Go? Remember, humans also use reinforcement learning in organizing actions. A reinforcement learning agent that has to optimize for a body that needs to drink, eat and socialize in order to function would totally go to grab a beer instead of playing a losing game. The needs of the agent are formative. Human needs are a source of much of our special skills. If we put artificial agents in similar situations, and they will be able to do similar things to humans.
Holy Shit! Donald Trump is a Slashdotter! Don't worry Donnie. They'll be working on bigger hands next!
Guns don't kill people; Physics kills people! - John Lithgow as Dick Solomon on Third Rock From The Sun
I normally try to read the whole article before commenting, but it starts with a list of straw men claims, so I didn't bother.
1. Artificial intelligence is already getting smarter than us, at an exponential rate.
It would be more accurate to say that the claim is that Artificial Intelligence is increasing faster than ours, which is hard to dispute. Saying "getting smarter" makes it sound like a claim that AI is already smarter, which I don't think anyone has made.
2. We’ll make AIs into a general purpose intelligence, like our own.
Why not? The ability to learn to play Go and Poker better than humans, without having detailed algorithms built in, shows that computational brute force goes a long way, even when humans don't understand how the program works. Until recently it was thought that there would have to be conceptual advances specific to those games in order to defeat human champions (and in any case it was already possible to defeat the average human).
3. We can make human intelligence in silicon.
It's unnecessary for AI to emulate human intelligence (and chauvinistic to suggest that it has to). Its capabilities can match or exceed humans, while working in a completely different way.
4. Intelligence can be expanded without limit.
Why? All that's necessary is for AI to equal or exceed human capabilities. Even if one makes the farfetched assumption that humans are at the peak of intelligence, simply being able to match the most intelligent humans would exceed the capabilities of 99.9+% of the population.
5. Once we have exploding superintelligence it can solve most of our problems.
It would probably allow solving most of our existing problems, and create new ones. Life goes on. In any case what it could accomplish is completely independent of whether it's possible.
Human (or just vision in general) is the best example. It accounts for 30% of the brain capacity. At one end, you have the human eye with a retina consisting of 100 million rods and cones. Then just in that space of a 5mm disc, there are seven layers of processing used to do contrast detection between colors and intensity along with edge detection. The optic nerve takes the compressed information from a thousand areas then passes it through to the brain into two paths; one to identify where objects are, the other to determine what the objects are and their orientation. Understanding what just a single region or layer of brain cells does leads to dozens of papers being published and advances in digital photography (image stabilization, motion correction).
Vintage computer adverts: http://www.vintageadbrowser.com/computers-and-software-ads
This seems to be an example of some kind of unbounded technological/scientific optimism that disregards the fact that during that history you're using as proof, we have also refined an understanding of physical limits that have not fundamentally changed. Think about laws of thermodynamics or the speed of light as a hard limit, among other things. We are not getting around those any time soon.
Of course if you're counting on a complete revolution of Physics, you're going to need "extraordinary evidence" to overturn a lot of what we already know. This is a tall order; even the theory of relativity and quantum mechanics do not do things like totally overturn Newton's ideas in our everyday life. You can't just expect these kinds of things to happen.
Then there is just some weirdness in the post...
What? The laws of physics have always had to be testable, otherwise you're just doing math. This is the reason the LHC was built, to be an experimental instrument. I do not understand the point about photons and gravitons; the former is a well-known quantum, the latter is theoretical. So far we haven't been able to quantize gravity.
Yeah, and time is a cube, eh?
No, limitations probably still are limitations, even when you develop a better understanding of what is going on. Stuff will fall down even tomorrow, even if you could demonstrate that you can quantize gravity. Getting around strongly established phenomena by better explanations would mean there is some until now completely non-observed part of the world we could exploit. This rarely happens so that what didn't work today, magically starts working tomorrow.
I want to play Free Market with a drowning Libertarian.
The Butlerian Jihad in the original Dune books was a reaction to the majority of humans delegating most of their thinking to machines, which allowed the humans that controlled the machines to control them. Every year, this seems more prophetic.
I am TheRaven on Soylent News
Indeed!
And one of the myricals in this is: if an object is about to hit your eyes or comes close by, the reflext to close the eyes and raise your hands etc. is triggered _before_ that information has even reached the brain/visual cortex.
The signal processing in the eye can bypass the visual cortex to trigger protective actions.
Cost free eBook I read (by iBook/Kobo/Amazon/ObookO/Gutenberg etc.): "The Green Odyssey" by Philip Jose Farmer.
1) AlphaGo exploded on the scene by beating world class Go players much faster and much earlier than expected. Exponentially is a loaded word. e^0.0000001 is an exponential growth rate. So let's not quibble about how exponential the growth rate is.
Particularly as we can't really measure intelligence. But "exponential" has a meaning, and it means a steady rate of doubling.
2) AlphaGo is a general purpose learning tool. Just listen to the lectures and articles penned by the DeepMind team.
No, it's a narrow AI. In the end, it's simply doing math. It's not "thinking" in any sense of the term. It's just able to hold many more probabilities in its memory than a human, and play them out much faster.
3) Alphago has displayed human-like intelligence, as claimed by the Go professionals it has played. They have said that AlphaGo plays like a human player.
That is what you call anthropomorphizing. The human players are simply projecting onto the machine.
4) If you take the fourth assumption literally, AI's intelligence is going to expand infinitely. Talking about infinity in human terms is unreasonable. Yes, AI's intelligence will expand.
"Infinite" simply means there's no limit. We don't know whether or not there's a limit to intelligence, but since the universe is finite, there would seem to be a limit to the things one could know. Infinite intelligence is our notion of God. What are the odds that infinite intelligence is also mythological?
-- sudon't
Air-ride Equipped
As my old friend Edsger Dijkstra once said "The question of whether Machines Can Think... is about as relevant as the question of whether Submarines Can Swim."
Given that our knowledge of the computational complexity of a single neuron is growing steadily, I think it's safe to say your FPGA cell estimate for a neuron was significantly too low. For example, scientists now know that one single neuron (of certain types) is an entire neural network all by itself. Dendrites with multiple localized spikes communicating with each other and with other cells. Ultimately performing non-linear computation prior to forwarding any signal to cell body.
Right you are. The absolute give-away (in addition to the ridiculous low-ball answer he provided) was "... that was pretty straightforward..." which shows the Dunning-Kruger Effect in full bloom. He had no idea now little he knows about the subject.
The example I like to use to illustrate how much smoke is being blown about this my tech types is the model organism Caenorhabditis elegans. This 1 mm long nematode has had every one of its 302 neurons in its nervous mapped out, including all connections to every other neuron, as well as the process of development from the initial fertilized egg - we have mapped out exactly how the nervous system develops (indeed every one of the 959 cells in its body have been similarly traced out).
Given this complete map of C. elegans nervous system we must have a spiffy computer of the little worm's "brain" able to replicate its behavior? Right?
Not even close. So far we cannot accurate model the behavior of even a single neuron in C. elegans. Even one single neuron represents computational complexity that we are still trying to understand.
Starships were meant to fly, Hands up and touch the sky - Nicky Minaj
On the other hand, neurons are severely limited by the biological processes, so it's possible that we can make artificial neurons that are better than the ones in our brains. A small neuron is 4 microns. A small transistor is 0.02 microns, so we can pack a lot of computation in the size of a neuron, and make it run millions of times faster too.
It is true that we can expect artificial neurons, once we know how to make one, will run much faster than natural ones, given the fact that we aren't limited to the materials that natural evolution must work with.
But the scale comparison you make (though a common one) is wildly, unjustly, favorable to current technology. The common "feature size" measure used to compare solid state circuit elements is in no way comparable to the computational units in nervous systems, which actually takes place at the level of individual molecules within a three dimensional neuron, part of a three dimensional closely packed neural structure.
That transistor is lying on (at this scale) an immensely thick slab of silicon, which we are trying to get down to 160 microns; which then gets stuffed into a gigantic package, which is mounted in a very space-inefficient way on a colossal board. So that the density of computational elements in a human-made system is actually many orders of magnitude lower than a biological system. Once you take into account the packaging of the highest gate count device currently on the market, the Stratix 10 FPGA, each of those 30 billion transistors occupies something 400 cubic microns, which isn't even considering the low density of package mounting in a complete computing system. Embedded in an actual computing system that volume grows to something like 10,000 cubic microns per transistor.
The rough (very rough) equivalent of a transistor in a natural neural system is not a neuron but a synapse, the behavior of which is still much more complex that a transistor. The average volume density of synapses in the human brain is about 0.1 cubic micron per synapse. If, for the sake of discussion, each synapse can be modeled with a 10 transistor gate array, then the effective density is one "transistor" per 0.01 cubic micron, or a million times smaller than those tiny transistor features we boast about. So our tiny transistors are "tiny" in only two dimensions, and then only if they are measured in isolation. In reality, compared to neurons, they are gigantic whales.
Starships were meant to fly, Hands up and touch the sky - Nicky Minaj