Re: Intelligence and Nomologicalism
On 24 Sep, 04:26, Rex Allen rexallen31...@gmail.com wrote: On Thu, Sep 23, 2010 at 12:12 PM, 1Z peterdjo...@yahoo.com wrote: On 22 Sep, 17:20, Rex Allen rexallen31...@gmail.com wrote: I guess I'd have to hear your definition of property to make any sense of that. In what sense is it like the properties of charge, mass, spin, or color? it's a distinguishing characteristic that is detectable So your position is that there is an algorithm that would correctly detect all instances of intelligence with no false positives? no. that isn't possible for physical properties either, and in any case has nothing to do with determinism If you possessed this algorithm, I could present you with a large cube of metal, silicon, and flashing lights, you could apply your algorithm to determine for certain whether any form of artificial intelligence was instantiated by the cube? No matter how obfuscated, encrypted, or abstract the representation used to instantiate the AI? This would be in contradiction to Hilary Putnam's work: Putnam's proposal, and its historical importance, was analyzed in detail in Piccinini forthcoming b. According to Putnam (1960, 1967, 1988), a system is a computing mechanism if and only if there is a mapping between a computational description and a physical description of the system. By computational description, Putnam means a formal description of the kind used in computability theory, such as a Turing Machine or a finite state automaton. Putnam puts no constraints on how to find the mapping between the computational and the physical description, allowing any computationally identified state to map onto any physically identified state. It is well known that Putnam's account entails that most physical systems implement most computations. This consequence of Putnam's proposal has been explicitly derived by Putnam (1988, pp. 95-96, 121-125) and Searle (1992, chap. 9). Or, as Hans Moravec puts it: What does it mean for a process to implement, or encode, a simulation? Something is palpably an encoding if there is a way of decoding or translating it into a recognizable form. Programs that produce pictures of evolving cloud cover from weather simulations, or cockpit views from flight simulations, are examples of such decodings. As the relationship between the elements inside the simulator and the external representation becomes more complicated, the decoding process may become impractically expensive. Yet there is no obvious cutoff point. A translation that is impractical today may be possible tomorrow given more powerful computers, some yet undiscovered mathematical approach, or perhaps an alien translator. Like people who dismiss speech and signs in unfamiliar foreign languages as meaningless gibberish, we are likely to be rudely surprised if we dismiss possible interpretations simply because we can't achieve them at the moment. Why not accept all mathematically possible decodings, regardless of present or future practicality? This seems a safe, open-minded approach, but it leads into strange territory. Where do you think that Putnam and Moravec went wrong? And in what sense is it different? it's not physically basic Then what is it? In what sense does it exist, if not physically? I assume your list of mass, charge, etc, were intended to be. Again, this has nothing to do with determinism Solving a problem correctly is no more impressive or significant than rain falling correctly. You answer the question in the only way the deterministic laws allow. The rain falls in the only way that the deterministic laws allow. so your actual conclusion is not that intelligence isn't intelligence, but that intelligence isn't an achivement No, my actual conclusion is the part where I conclude: The word 'intelligence' doesn't refer to anything except the experiential requirements that the universe places on you as a consequence of its causal structure. I have no idea what that means Okay, so here's a definition of intelligence from the Merriam-Webster dictionary: the ability to apply knowledge to manipulate one's environment or to think abstractly as measured by objective criteria (as tests) But what is an ability in a deterministic universe? It's something you can have, but not choose to have. It is not, in other words an achievement; one wuuld be no more responsible for ones rationality or intelligence than eye-colour For any given input, a deterministic system can only react in one way. If you expose a deterministic system to a set of inputs that represent a particular environment, the system will react in the one and only way it can to that set of inputs. Knowledge is just the internal state of the deterministic system. This is true of a human. This is true of a bacterium. This is true of a Roomba vacuum cleaner. This is true of a hurricane. This is true of a rock. ie they
Re: Intelligence and Nomologicalism
On 9/23/2010 8:26 PM, Rex Allen wrote: On Thu, Sep 23, 2010 at 12:12 PM, 1Zpeterdjo...@yahoo.com wrote: On 22 Sep, 17:20, Rex Allenrexallen31...@gmail.com wrote: I guess I'd have to hear your definition of property to make any sense of that. In what sense is it like the properties of charge, mass, spin, or color? it's a distinguishing characteristic that is detectable So your position is that there is an algorithm that would correctly detect all instances of intelligence with no false positives? If you possessed this algorithm, I could present you with a large cube of metal, silicon, and flashing lights, you could apply your algorithm to determine for certain whether any form of artificial intelligence was instantiated by the cube? No matter how obfuscated, encrypted, or abstract the representation used to instantiate the AI? This would be in contradiction to Hilary Putnam's work: Putnam's proposal, and its historical importance, was analyzed in detail in Piccinini forthcoming b. According to Putnam (1960, 1967, 1988), a system is a computing mechanism if and only if there is a mapping between a computational description and a physical description of the system. By computational description, Putnam means a formal description of the kind used in computability theory, such as a Turing Machine or a finite state automaton. Putnam puts no constraints on how to find the mapping between the computational and the physical description, allowing any computationally identified state to map onto any physically identified state. It is well known that Putnam's account entails that most physical systems implement most computations. This consequence of Putnam's proposal has been explicitly derived by Putnam (1988, pp. 95-96, 121-125) and Searle (1992, chap. 9). Or, as Hans Moravec puts it: What does it mean for a process to implement, or encode, a simulation? Something is palpably an encoding if there is a way of decoding or translating it into a recognizable form. Programs that produce pictures of evolving cloud cover from weather simulations, or cockpit views from flight simulations, are examples of such decodings. As the relationship between the elements inside the simulator and the external representation becomes more complicated, the decoding process may become impractically expensive. Yet there is no obvious cutoff point. A translation that is impractical today may be possible tomorrow given more powerful computers, some yet undiscovered mathematical approach, or perhaps an alien translator. Like people who dismiss speech and signs in unfamiliar foreign languages as meaningless gibberish, we are likely to be rudely surprised if we dismiss possible interpretations simply because we can't achieve them at the moment. Why not accept all mathematically possible decodings, regardless of present or future practicality? This seems a safe, open-minded approach, but it leads into strange territory. Where do you think that Putnam and Moravec went wrong? And in what sense is it different? it's not physically basic Then what is it? In what sense does it exist, if not physically? Solving a problem correctly is no more impressive or significant than rain falling correctly. You answer the question in the only way the deterministic laws allow. The rain falls in the only way that the deterministic laws allow. so your actual conclusion is not that intelligence isn't intelligence, but that intelligence isn't an achivement No, my actual conclusion is the part where I conclude: The word 'intelligence' doesn't refer to anything except the experiential requirements that the universe places on you as a consequence of its causal structure. I have no idea what that means Okay, so here's a definition of intelligence from the Merriam-Webster dictionary: the ability to apply knowledge to manipulate one's environment or to think abstractly as measured by objective criteria (as tests) But what is an ability in a deterministic universe? For any given input, a deterministic system can only react in one way. If you expose a deterministic system to a set of inputs that represent a particular environment, the system will react in the one and only way it can to that set of inputs. And if that reaction is to manipulate it's envrionment is a way advantageous to it, it's intelligent. Intelligence must always be relative to some situation or environment. That's where Putnam and Moravec go wrong and Merriam-Webster get it right. Knowledge is just the internal state of the deterministic system. That's not a usable definition: internal=inaccessible. Knowledge must be expressible. It must be information that makes a difference. Otherwise you fall into the paradox of the rock that computes everything. Brent This is true of a human. This is true of a bacterium. This is true of a Roomba vacuum cleaner.