jeff@aiai.ed.ac.uk (Jeff Dalton) (03/17/89)
In article <1989Mar4.152943.10902@cs.rochester.edu> yamauchi@cs.rochester.edu (Brian Yamauchi) writes: >I agree. Of course, the same thing applies to robots. Suppose an >humanoid robot walking next to you stubbed his toe and said "That >hurts!". Would you respond with "No, you're just programmed to say >that when you damage yourself!" Sure. Why not? Maybe I happen to know that the robot *is* programmed to say that. After all, all you've said about the robot is that it's humanoid and that it says "that hurts" when it stubs its toe. But suppose you tell me lots more about the robot. Maybe it passes the LTT and the TTT, and so on. However, if it does pass the various Turing Tests, then it follows that you can't tell me all that much about it's programming -- because right now we have essentially no idea how to program anything that can pass those tests. Maybe, when we know a lot more about such robots (assuming they're possible at all) and about ourselves, it will be pretty clear that that robot does feel pain. But maybe it will instead be clear that it's just programmed to behave as if it feels pain. How do you know it won't turn out that way?
bwk@mbunix.mitre.org (Barry W. Kort) (03/18/89)
In article <337@skye.ed.ac.uk> jeff@aiai.UUCP (Jeff Dalton) writes: > Maybe, when we know a lot more about such robots (assuming they're > possible at all) and about ourselves, it will be pretty clear that > that robot does feel pain. But maybe it will instead be clear that > it's just programmed to behave as if it feels pain. How do you know > it won't turn out that way? Perhaps it would help if we define pain. The robot has a mechanical body, which it relies on for locomotion and interaction with the environmental surround. Preservation of the functional integrity of that mechanical body is a priority if the robot is to pursue goals which depend on that body. Therefore, the robot needs to be informed of any danger or damage to its pieceparts. Mechanical stress sensors report such information, and the robot can act to moderate or ameliorate conditions which endanger it's long-term corporeal well-being. In humans, we call such information "pain". In robots we can choose to call it what we like, but functionally it serves the same purpose. --Barry Kort
jack@cs.glasgow.ac.uk (Jack Campin) (03/22/89)
bwk@mbunix (Barry Kort) wrote: Perhaps it would help if we define pain [...] the robot needs to be informed of any danger or damage to its pieceparts. Mechanical stress sensors report such information, and the robot can act to moderate or ameliorate conditions which endanger it's long-term corporeal well-being. In humans, we call such information "pain". In robots we can choose to call it what we like, but functionally it serves the same purpose. This definition doesn't work. If, while under spinal block anaesthesia, I see a rat starting to chew my toe off, I'm getting information about potential damage to my body, but that information does not constitute "pain". -- Jack Campin * Computing Science Department, Glasgow University, 17 Lilybank Gardens, Glasgow G12 8QQ, SCOTLAND. 041 339 8855 x6045 wk 041 556 1878 ho INTERNET: jack%cs.glasgow.ac.uk@nss.cs.ucl.ac.uk USENET: jack@glasgow.uucp JANET: jack@uk.ac.glasgow.cs PLINGnet: ...mcvax!ukc!cs.glasgow.ac.uk!jack