1 00:00:04,280 --> 00:00:09,111 While general game playing is a topic within here and interest, work is in this 2 00:00:09,111 --> 00:00:13,759 area has practical value as well. The underlying technology can be used in a 3 00:00:13,759 --> 00:00:18,835 variety of other application areas, such as enterprise management and computational 4 00:00:18,835 --> 00:00:25,506 law. Science, engineering, and so forth. In the case of enterprise management, 5 00:00:25,506 --> 00:00:32,331 computational law, just consider that the rules of games are analogous to the laws 6 00:00:32,331 --> 00:00:40,041 of society, the rules of business, and so forth. More fundamentally, general game 7 00:00:40,041 --> 00:00:43,799 playing is valued as a test bed for theories of intelligence. Game 8 00:00:43,799 --> 00:00:48,013 descriptions provide full information about a world and determine optimal 9 00:00:48,013 --> 00:00:52,226 strategies. It is a base line for evaluating the behavior of actual agents. 10 00:00:52,226 --> 00:00:57,009 By its nature the GP setting can be used to evaluate problem solving strategies and 11 00:00:57,009 --> 00:01:01,336 by extension theories of intelligence by taking into account representation, 12 00:01:01,336 --> 00:01:07,123 incompleteness of information, and resource balance. That said, it's worth 13 00:01:07,123 --> 00:01:13,240 noting that there are some shortcomings of the current state of GDP as a test bed for 14 00:01:13,240 --> 00:01:17,844 AI. First, there's the tabula rasa critique. It goes, as follows, human 15 00:01:17,844 --> 00:01:23,343 intelligence is arguably the product of eons of evolution. We are to some extent, 16 00:01:23,343 --> 00:01:28,841 at least, wired to function well in this world. General game players have nothing 17 00:01:28,841 --> 00:01:34,013 at their disposal but pure mathematics. Okay. That's true. However a key indicator 18 00:01:34,013 --> 00:01:38,370 of intelligence is the ability of each individual to function in radically new 19 00:01:38,370 --> 00:01:42,782 environments. And general game playing is exploring that capability, that play of 20 00:01:42,782 --> 00:01:49,932 our intelligence. Second shortcoming to GDP, is the learning critique. In GGP, 21 00:01:49,932 --> 00:01:54,470 players are told how the world works. The critiquos, that real intelligence requires 22 00:01:54,470 --> 00:01:57,970 the ability to figure out new environments. Okay, well there's no 23 00:01:57,970 --> 00:02:02,289 question that the ability to learn is essential. However, real intelligence also 24 00:02:02,289 --> 00:02:06,718 requires the ability to use theories once they're formed. And that's what general 25 00:02:06,718 --> 00:02:10,818 game playing is emphasizing. This is really the mean of what's often called 26 00:02:10,818 --> 00:02:15,083 general systems, or universal systems, or sometimes declarative systems. A nd the 27 00:02:15,083 --> 00:02:19,546 interest in this problem dates back to the very beginnings of the field. It was in 28 00:02:19,546 --> 00:02:23,655 1958 that John McCarthy invented the concept of the advice taker. The idea was 29 00:02:23,655 --> 00:02:27,869 simple, he wanted a machine that he could program by description. He would scribe 30 00:02:27,869 --> 00:02:31,925 the intended environment and the desired goal, and the machine would use that 31 00:02:31,925 --> 00:02:36,191 information in determining it's behavior. There would be no programming in the 32 00:02:36,191 --> 00:02:40,575 traditional sense. McCarthy presented this concept in a paper that has become a 33 00:02:40,575 --> 00:02:44,923 classic in the field of AI. and let me just read this quote. The main advantage 34 00:02:44,923 --> 00:02:49,486 we expect the advice taker to have is that its behavior will be improvable merely by 35 00:02:49,486 --> 00:02:53,727 making statements to it. Telling it about its environment, and what it is wanted 36 00:02:53,727 --> 00:02:57,861 from it. To make these statements will require little if any knowledge of the 37 00:02:57,861 --> 00:03:02,989 program. Or the previous knowledge of the advice taker. Well that's an ambitious 38 00:03:02,989 --> 00:03:08,266 goal. But that was a time of high hopes and grand ambitions. The idea caught the 39 00:03:08,266 --> 00:03:13,543 imaginations of numerous subsequent researchers, notably Bob Kowalsky, one of 40 00:03:13,543 --> 00:03:18,218 the high priests of logic programming. And then Feigenbaum, the inventor of knowledge 41 00:03:18,218 --> 00:03:23,387 engineering. in a paper written in 1974, Feigenbaum gave his most forceful 42 00:03:23,387 --> 00:03:28,778 statement of McCarthy's ideal. He says, the potential use of computers by people 43 00:03:28,778 --> 00:03:33,077 to accomplish tasks, can be one-dimensionalized into a spectrum, 44 00:03:33,077 --> 00:03:38,605 representing the nature of instruction that must be given the computer to do its 45 00:03:38,605 --> 00:03:43,427 job. Call it the what-to-how spectrum. At one extreme of the spectrum, the user 46 00:03:43,427 --> 00:03:48,196 supplies his intelligence to instruct the machine with precision exactly how to do 47 00:03:48,196 --> 00:03:52,792 his job step by step. At the other end of the spectrum is the user with his real 48 00:03:52,792 --> 00:03:57,502 problem. He aspires to communicate what he wants done, without having to lay out in 49 00:03:57,502 --> 00:04:02,058 detail all necessary sub-goals for adequate performance. Okay, one final 50 00:04:02,058 --> 00:04:07,365 remark. Some have argued that, the way to achieve intelligent behavior is through 51 00:04:07,365 --> 00:04:12,671 specialization. That may work, so long as the assumptions one makes in building such 52 00:04:12,671 --> 00:04:17,210 systems are true. For general intellige nce, however, general intellectual 53 00:04:17,210 --> 00:04:22,517 capabilities are needed. And such systems should be capable of performing well in a 54 00:04:22,517 --> 00:04:27,567 wide variety of tasks. To paraphrase the words of Robert Heinlein, a human being, 55 00:04:27,567 --> 00:04:32,925 or a computer robot, should be able to change a diaper. Plan an invasion, butcher 56 00:04:32,925 --> 00:04:38,640 a hog, con a ship. Design a building, write a sonnet, balance accounts. Build a 57 00:04:38,640 --> 00:04:44,584 wall, set a bone. Comfort the dying. Take orders, give orders. Cooperate. Act alone. 58 00:04:44,584 --> 00:04:50,452 Solve equations. Analyze a new problem. Pitch manure. Program a computer. Cook a 59 00:04:50,452 --> 00:04:56,015 tasty meal. Fight efficiently, die gallantly. Specialization is for insects. 60 00:04:56,015 --> 00:05:02,036 Those of us who are more interested in artificial intelligence than artificial 61 00:05:02,036 --> 00:05:04,170 insects agree with Heinlein.