There is not a fixed amount of jobs in any economy. The demand for labor is never satiated, because the demand for economic goods, by definition, is never satiated. The computer replaced the typewriter and yet we managed to avoid systemic and extended structural unemployment. This is because the introduction of new technology actually introduces new opportunities and frees up labor for other, more warranted economic activities. Those who produced typewriters found employment, and whole new markets opened up for computers and the inputs required to produce computers. We needed computer engineers, manufactures, the capital required to build computers (semiconductors, motherboards, ect), ect ect.
This is an ancient Marxian fallacy/dogma. There is no struggle between capital and labor; they are co-dependent factors of production. All economic phenomena is a positive sum game.
What Ultimate is proposing is a society in which robots are basically human, except for specific biological differences that would set them apart. But, he believes that not only will they be fully able to mimick the human brain, but that that brain would act rationally and would enjoy the same lust for entrepreneurship. This effectively means that humankind would have created a competitor “species” (poor word choice, maybe, since it’s not biological; but, you know what I mean).
This sounds like something straight out of Terminator or the Matrix.
Absolutely. People have more time for leisure because capital accumulation (and capital goods) have lead to increases in productivity. Higher (real) wages means less hours need to be worked to maintain the same lifestyle, or in some cases, maintain a somewhat steady increase in standard of living.
I’m always reminded of this article by Lew Rockwell about how far we have come in a relatively short time with some economic freedom and minimal government.
Then you haven’t been understanding my point, either. We can simulate physical systems now, just not systems of sufficient complexity to handle all of the chemical and physical interactions that go on inside a human brain. When it comes to simulating the brain, there are only incremental steps between here and there. We know that it is doable given a certain amount of computing power, since we already know that it is possible to simulate the laws of physics.
Your first two sentences are unbacked statements veiled as facts. Simply because we don’t understand how the brain processes information does not mean that it does not process information. Why do you believe that chemical and physical interactions cannot be simulated? Why do you think that not only will we never be able to do such a simulation, but furthermore, that we will never be able to use such simulations to further our understanding of brain processes and further our understanding of how the brain works?
Well, I respectfully disagree with you. You can only consistently believe this if you believe that the human brain is not an entirely physical process or that technological progress will come to a permanent standstill. Otherwise, there is no reason that it cannot be simulated, and there is also no reason that we cannot learn how those processes work or learn how to reverse engineer it and synthesize an artificially-constructed brain other than your belief that it will never be possible, even across an infinite span of time.
Yeah, I see that, but it’s just an angels-dancing-on-a-pin argument. It’s like, “What if there were humans… but they weren’t really humans? What then?”
What’s also interesting is if you can find ways to accelerate live human brains via implants or other means.
Except that computers aren’t smart; they have to be programmed to do all of these things. Where things get really interesting if once you start putting in some general intelligence. That’s the hard problem that still needs to be solved, but it’s always possible to brute-force it simply by taking what already exists in our heads and observing it at the finest level of details. That kind of resolution is still a long ways away, of course, which is why researchers are trying to find smarter ways of doing this.
I only agree that computers, by their very nature, can only solve what is computable; but then again, if the human brain obeys the laws of physics and there is nothing “special” about it, then the same limitation applies to it as well. With that in mind, then I agree.
Then you haven’t been understanding my point, either. We can simulate physical systems now, just not systems of sufficient complexity to handle all of the chemical and physical interactions that go on inside a human brain. When it comes to simulating the brain, there are only incremental steps between here and there. We know that it is doable given a certain amount of computing power, since we already know that it is possible to simulate the laws of physics.
The problem is that you paint these interactions between chemicals and what have you as easily recognizable patterns that occur over and over again. This is an assumption, and may not represent reality. The truth is that these chemical reactions may depend entirely on how the brain interprets certain events, and that one chemical reaction can be very different to another, depending on your sensory capabilities and what not. This may be why there is such wide variance in regards to how people think.
You keep mentioning the laws of physics, but these may have nothing to do with the subject being discussed.
Your first two sentences are unbacked statements veiled as facts. Simply because we don’t understand how the brain processes information does not mean that it does not process information.
I never said anything of the like.
Why do you believe that chemical and physical interactions cannot be simulated? Why do you think that not only will we never be able to do such a simulation, but furthermore, that we will never be able to use such simulations to further our understanding of brain processes and further our understanding of how the brain works?
See above.
You can only consistently believe this if you believe that the human brain is not an entirely physical process or that technological progress will come to a permanent standstill. Otherwise, there is no reason that it cannot be simulated, and there is also no reason that we cannot learn how those processes work or learn how to reverse engineer it and synthesize an artificially-constructed brain other than your belief that it will never be possible, even across an infinite span of time.
This is a false dichotomy. I can also believe that there is no single or clear pattern how the brain works, and that the method by which it works also can develop over the timespan of a single human life.
Yeah, I see that, but it’s just an angels-dancing-on-a-pin argument. It’s like, “What if there were humans… but they weren’t really humans? What then?”
Clayton -
Why is why, ultimately, this argument is not going to go anywhere.
Like I said, this is a similar to someone trying to prove God exists. Maybe not as radical, or difficult to prove eitherway, but probably just as implausible.
The honorable Sir Roger Penrose to the contrary, I think our brains only solve computable problems. I doubt we would struggle with the laws of Einsteinian physics or quantum mechincs - or both together - if we could solve uncomputable problems. The laws of physics - indeed, even all of mathematics - would be easier than tic-tac-toe, if that were the case. If we could solve the halting problem, for example, we could simply write a computer program that says, “If theorem X is true, halt, else, don’t halt” and then decide whether that program halts or not. Alas, mathematics is a hell of a lot harder than that, for our pitiful brains.
The easy way to see that we cannot build computers that can solve problems which our brains could not, in principle, solve, is to note that it is a hidden contradiction! If our brains built the computer that goes on to solve the supposedly “unsolvable” problem, then our brains have, in fact, solved the problem! Now, you can get into trouble with this purely metaphysical argument but it points the way to a more rigorously stated result in mathematics that traces its roots ultimately back to Godel. You can read about it here. (The relevant phrase from the paper is this, “N bits of axioms are needed to be able to prove that an N-bit program is elegant” which through a rather circuitous argument can be translated into saying that the brain - if it is a Turing machine - cannot find its own ultimate explanation or, if you will, its own shortest program… I’m thinking of submitting an article to Mises.org on this subject since it touches on Mises’s calculation argument).
Brain’s much worse than a Turing Machine. A turing machine has infinite tape and infinite time. If our brains were turing machines then who would care about the halting problem at all?
Anyway the entire “robots will take over and we won’t have any jobs” thing reeks. Ever heard of the law of comparative advantage? Two acting beings can ALWAYS gain by cooperating, even if one is absolutely superior to the other in every way.
Easy, if you assume sufficiently powerful computers. Make a simulation of the Earth, save state, then give everyone on this simulated planet a randomly generated sequence of words of a random, novel-like length, translate into every language, and measure its popularity. Save this word sequence and record its probability, then restore the previous state of the Earth simulation. Repeat Graham’s number times, ensuring each word sequence is unique, and you’ll have the most popular novel ever written. You can vastly reduce the number of repeats needed by introducing heuristics ahead of time to ensure the sentences are coherent and the like.
That said there’s probably a much easier way, but it’s definitely possible, at least hypothetically. Whether it’s physically possible to make a computer that powerful is another story, but like I said, probably a much less costly algorithm.
Easy, if you assume sufficiently powerful computers. Make a simulation of the Earth, save state, then give everyone on this simulated planet a randomly generated sequence of words of a random, novel-like length, translate into every language, and measure its popularity. Save this word sequence and record its probability, then restore the previous state of the Earth simulation. Repeat Graham’s number times, ensuring each word sequence is unique, and you’ll have the most popular novel ever written. You can vastly reduce the number of repeats needed by introducing heuristics ahead of time to ensure the sentences are coherent and the like.
This might not even be enough. What makes certain novels great, or music, or similar things, is not that they appeal to existing tastes, rather that they are able to successfully break barriers in taste. If works of art were developed around existing tastes, we would have never had changes in genre (which, in most cases, were at first very unpopular). This requires innovation.
Easy. Just expand the popularity test to every human being who’s ever lived, or ever will live in the future (given the constraints of the simulation, perfect foreknowledge is uncomputable.) It wouldn’t be perfect but it’d be much better than any human could possibly pull off.
That said the “elegant” solution probably won’t be all that hard, given our abilities to do things like write novels has only recently evolved and that part of our brain is very poorly developed. It’ll almost certainly be easier than beating humans at image recognition, for instance.
I don’t really see the problem. A similar argument can be held if you consider, hypothetically, if 2 sentient species, about equally intelligent overall, but of radically different evolutionary lineage, came to inhabit a planet together. Would they, by definition, always be at odds with each other? I see no reason at all to believe the law of comparative advantage holds differently over species than it does over ethnicity. Why should it not hold over biologies too?
Easy. Just expand the popularity test to every human being who’s ever lived, or ever will live in the future (given the constraints of the simulation, perfect foreknowledge is uncomputable.) It wouldn’t be perfect but it’d be much better than any human could possibly pull off.
I’m not sure how that addresses the notion of innovation, and writing something that may not be tasteful at one point, but becomes tasteful after it’s written.
You can only consistently believe this if you believe that the human brain is not an entirely physical process or that technological progress will come to a permanent standstill. Otherwise, there is no reason that it cannot be simulated, and there is also no reason that we cannot learn how those processes work or learn how to reverse engineer it and synthesize an artificially-constructed brain other than your belief that it will never be possible, even across an infinite span of time.