Re: [agi] How Would You Design a Play Machine?
Mike Tintner wrote: ...how would you design a play machine - a machine that can play around as a child does? I wouldn't. IMHO that's just another waste of time and effort (unless it's being done purely for research purposes). It's a diversion of intellectual and financial resources that those serious about building an AGI any time in this century cannot afford. I firmly believe if we had not set ourselves the goal of developing human-style intelligence (embodied or not) fifty years ago, we would already have a working, non-embodied AGI. Turing was wrong (or at least he was wrongly interpreted). Those who extended his imitation test to humanoid, embodied AI were even more wrong. We *do not need embodiment* to be able to build a powerful AGI that can be of immense utility to humanity while also surpassing human intelligence in many ways. To be sure, we want that AGI to be empathetic with human intelligence, but we do not need to make it equivalent (i.e., just like us). I don't want to give the impression that a non-Turing intelligence will be easy to design and build. It will probably require at least another twenty years of two steps forward, one step back effort. So, if we are going to develop a non-human-like, non-embodied AGI within the first quarter of this century, we are going to have to just say no to Turing and start to use human intelligence as an inspiration, not a destination. Cheers, Brad Mike Tintner wrote: Just a v. rough, first thought. An essential requirement of an AGI is surely that it must be able to play - so how would you design a play machine - a machine that can play around as a child does? You can rewrite the brief as you choose, but my first thoughts are - it should be able to play with a) bricks b)plasticine c) handkerchiefs/ shawls d) toys [whose function it doesn't know] and e) draw. Something that should be soon obvious is that a robot will be vastly more flexible than a computer, but if you want to do it all on computer, fine. How will it play - manipulate things every which way? What will be the criteria of learning - of having done something interesting? How do infants, IOW, play? --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] Re: I Made a Mistake
Chill down Jim, he took it back. On 8/24/08, Jim Bromer [EMAIL PROTECTED] wrote: Intolerance of another person's ideas through intimidation or ridicule is intellectual repression. You won't elevate a discussion by promoting a program anti-intellectual repression. Intolerance of a person for his religious beliefs is a form of intellectual intolerance --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] The Necessity of Embodiment
In other words, Vladimir, you are suggesting that an AGI must be at some level controlled from humans, therefore not 'fully-embodied' in order to prevent non-friendly AGI as the outcome. Therefore humans must somehow be able to control its goals, correct? Now, what if controlling those goals would entail not being able to create an AGI, would you suggest we should not create one, in order to avoid the disastrous consequences you mentioned? Valentina --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] rpi.edu
Eric, http://www.cogsci.rpi.edu/research/rair/asc_rca/ Sorry, couldn't answer your question based on quick read. Cheers, Brad Eric Burton wrote: Does anyone know if Rensselaer Institute is still on track to crack the Turing Test by 2009? There was a Slashdot article or two about their software called 'RASCALS' earlier this year. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] The Necessity of Embodiment
On Mon, Aug 25, 2008 at 1:07 PM, Valentina Poletti [EMAIL PROTECTED] wrote: In other words, Vladimir, you are suggesting that an AGI must be at some level controlled from humans, therefore not 'fully-embodied' in order to prevent non-friendly AGI as the outcome. Controlled in Friendliness sense of the word. (I still have no idea what embodied refers to, now that you, me and Terren used it in different senses, and I recall reading a paper about 6 different meanings of this word in academic literature, none of them very useful). Therefore humans must somehow be able to control its goals, correct? Now, what if controlling those goals would entail not being able to create an AGI, would you suggest we should not create one, in order to avoid the disastrous consequences you mentioned? Why would anyone suggest creating a disaster, as you pose the question? -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.com/ --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Brad, That's sad. The suggestion is for a mental exercise, not a full-scale project. And play is fundamental to the human mind-and-body - it characterises our more mental as well as more physical activities - drawing, designing, scripting, humming and singing scat in the bath, dreaming/daydreaming much more. It is generally acknowledged by psychologists to be an essential dimension of creativity - which is the goal of AGI. It is also an essential dimension of animal behaviour and animal evolution. Many of the smartest companies have their play areas. But I'm not aware of any program or computer design for play - as distinct from elaborating systematically and methodically or genetically on themes - are you? In which case it would be good to think about one - it'll open your mind give you new perspectives. This should be a group where people are not too frightened to play around with ideas. Brad: Mike Tintner wrote: ...how would you design a play machine - a machine that can play around as a child does? I wouldn't. IMHO that's just another waste of time and effort (unless it's being done purely for research purposes). It's a diversion of intellectual and financial resources that those serious about building an AGI any time in this century cannot afford. I firmly believe if we had not set ourselves the goal of developing human-style intelligence (embodied or not) fifty years ago, we would already have a working, non-embodied AGI. Turing was wrong (or at least he was wrongly interpreted). Those who extended his imitation test to humanoid, embodied AI were even more wrong. We *do not need embodiment* to be able to build a powerful AGI that can be of immense utility to humanity while also surpassing human intelligence in many ways. To be sure, we want that AGI to be empathetic with human intelligence, but we do not need to make it equivalent (i.e., just like us). I don't want to give the impression that a non-Turing intelligence will be easy to design and build. It will probably require at least another twenty years of two steps forward, one step back effort. So, if we are going to develop a non-human-like, non-embodied AGI within the first quarter of this century, we are going to have to just say no to Turing and start to use human intelligence as an inspiration, not a destination. Cheers, Brad Mike Tintner wrote: Just a v. rough, first thought. An essential requirement of an AGI is surely that it must be able to play - so how would you design a play machine - a machine that can play around as a child does? You can rewrite the brief as you choose, but my first thoughts are - it should be able to play with a) bricks b)plasticine c) handkerchiefs/ shawls d) toys [whose function it doesn't know] and e) draw. Something that should be soon obvious is that a robot will be vastly more flexible than a computer, but if you want to do it all on computer, fine. How will it play - manipulate things every which way? What will be the criteria of learning - of having done something interesting? How do infants, IOW, play? --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Kittens play with small moving objects because it teaches them to be better hunters. Play is not a goal in itself, but a subgoal that may or may not be a useful part of a successful AGI design. -- Matt Mahoney, [EMAIL PROTECTED] - Original Message From: Mike Tintner [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Monday, August 25, 2008 8:59:06 AM Subject: Re: [agi] How Would You Design a Play Machine? Brad, That's sad. The suggestion is for a mental exercise, not a full-scale project. And play is fundamental to the human mind-and-body - it characterises our more mental as well as more physical activities - drawing, designing, scripting, humming and singing scat in the bath, dreaming/daydreaming much more. It is generally acknowledged by psychologists to be an essential dimension of creativity - which is the goal of AGI. It is also an essential dimension of animal behaviour and animal evolution. Many of the smartest companies have their play areas. But I'm not aware of any program or computer design for play - as distinct from elaborating systematically and methodically or genetically on themes - are you? In which case it would be good to think about one - it'll open your mind give you new perspectives. This should be a group where people are not too frightened to play around with ideas. Brad: Mike Tintner wrote: ...how would you design a play machine - a machine that can play around as a child does? I wouldn't. IMHO that's just another waste of time and effort (unless it's being done purely for research purposes). It's a diversion of intellectual and financial resources that those serious about building an AGI any time in this century cannot afford. I firmly believe if we had not set ourselves the goal of developing human-style intelligence (embodied or not) fifty years ago, we would already have a working, non-embodied AGI. Turing was wrong (or at least he was wrongly interpreted). Those who extended his imitation test to humanoid, embodied AI were even more wrong. We *do not need embodiment* to be able to build a powerful AGI that can be of immense utility to humanity while also surpassing human intelligence in many ways. To be sure, we want that AGI to be empathetic with human intelligence, but we do not need to make it equivalent (i.e., just like us). I don't want to give the impression that a non-Turing intelligence will be easy to design and build. It will probably require at least another twenty years of two steps forward, one step back effort. So, if we are going to develop a non-human-like, non-embodied AGI within the first quarter of this century, we are going to have to just say no to Turing and start to use human intelligence as an inspiration, not a destination. Cheers, Brad Mike Tintner wrote: Just a v. rough, first thought. An essential requirement of an AGI is surely that it must be able to play - so how would you design a play machine - a machine that can play around as a child does? You can rewrite the brief as you choose, but my first thoughts are - it should be able to play with a) bricks b)plasticine c) handkerchiefs/ shawls d) toys [whose function it doesn't know] and e) draw. Something that should be soon obvious is that a robot will be vastly more flexible than a computer, but if you want to do it all on computer, fine. How will it play - manipulate things every which way? What will be the criteria of learning - of having done something interesting? How do infants, IOW, play? --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] The Necessity of Embodiment
On 8/25/08, Vladimir Nesov [EMAIL PROTECTED] wrote: On Mon, Aug 25, 2008 at 1:07 PM, Valentina Poletti [EMAIL PROTECTED] wrote: In other words, Vladimir, you are suggesting that an AGI must be at some level controlled from humans, therefore not 'fully-embodied' in order to prevent non-friendly AGI as the outcome. Controlled in Friendliness sense of the word. (I still have no idea what embodied refers to, now that you, me and Terren used it in different senses, and I recall reading a paper about 6 different meanings of this word in academic literature, none of them very useful). Agree Therefore humans must somehow be able to control its goals, correct? Now, what if controlling those goals would entail not being able to create an AGI, would you suggest we should not create one, in order to avoid the disastrous consequences you mentioned? Why would anyone suggest creating a disaster, as you pose the question Also agree. As far as you know, has anyone, including Eliezer, suggested any method or approach (as theoretical or complicated as it may be) to solve this problem? I'm asking this because the Singularity has confidence in creating a self-improving AGI in the next few decades, and, assuming they have no intention to create the above mentioned disaster.. I figure someone must have figured some way to approach this problem. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: Information theoretic approaches to AGI (was Re: [agi] The Necessity of Embodiment)
John, I have looked at your patent and various web pages. You list a lot of nice sounding ethical terms (honor, love, hope, peace, etc) but give no details on how to implement them. You have already admitted that you have no experimental results, haven't actually built anything, and have no other results such as refereed conference or journal papers describing your system. If I am wrong about this, please let me know. -- Matt Mahoney, [EMAIL PROTECTED] - Original Message From: John LaMuth [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, August 24, 2008 11:21:30 PM Subject: Re: Information theoretic approaches to AGI (was Re: [agi] The Necessity of Embodiment) - Original Message - From: Matt Mahoney [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Sunday, August 24, 2008 2:46 PM Subject: Re: Information theoretic approaches to AGI (was Re: [agi] The Necessity of Embodiment) I have challenged this list as well as the singularity and SL4 lists to come up with an example of a mathematical, software, biological, or physical example of RSI, or at least a plausible argument that one could be created, and nobody has. To qualify, an agent has to modify itself or create a more intelligent copy of itself according to an intelligence test chosen by the original. The following are not examples of RSI: 1. Evolution of life, including humans. 2. Emergence of language, culture, writing, communication technology, and computers. -- Matt Mahoney, [EMAIL PROTECTED] ### * Matt Where have you been for the last 2 months ?? I had been talking then about my 2 US Patents for ethical/friendly AI along lines of a recursive simulation targeting language (topic 2) above. This language agent employs feedback loops and LTM to increase comprehension and accuracy (and BTW - resolves the ethical safeguard problems for AI) ... No-one yet has proven me wrong ?? Howsabout YOU ??? More at www.angelfire.com/rnb/fairhaven/specs.html John LaMuth www.ethicalvalues.com agi | Archives | Modify Your Subscription --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Matt: Kittens play with small moving objects because it teaches them to be better hunters. Play is not a goal in itself, but a subgoal that may or may not be a useful part of a successful AGI design. Certainly, crude imitation of, and preparation for, adult activities is one aspect of play. But pure exploration - experimentation -and embroidery also are important. An infant dropping throwing things handling things every which way. Doodling - creating lines that go off and twist and turn in every direction. Babbling - playing around with sounds. Sputtering - playing around with silly noises - kids love that, no? (Even some of us adults too). Playing with stories and events - and alternative endings, beginnings and middles. Make believe. Playing around with the rules of invented games. Human development allots a great deal of time for such play. --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] The Necessity of Embodiment
On Mon, Aug 25, 2008 at 6:23 PM, Valentina Poletti [EMAIL PROTECTED] wrote: On 8/25/08, Vladimir Nesov [EMAIL PROTECTED] wrote: Why would anyone suggest creating a disaster, as you pose the question? Also agree. As far as you know, has anyone, including Eliezer, suggested any method or approach (as theoretical or complicated as it may be) to solve this problem? I'm asking this because the Singularity has confidence in creating a self-improving AGI in the next few decades, and, assuming they have no intention to create the above mentioned disaster.. I figure someone must have figured some way to approach this problem. I see no realistic alternative (as in with high probability of occurring in actual future) to creating a Friendly AI. If we don't, we are likely doomed one way or another, most thoroughly through Unfriendly AI. As I mentioned, one way to see Friendly AI is as a second chance substrate, which is a first thing to do to ensure any kind of safety from fatal or just vanilla bad mistakes in the future. Of course, establishing a dynamics that know a mistake and when to recover or prevent or guide is a tricky part. -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.com/ --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Actually, kittens play because it's fun. Evolution has equipped them with the rewarding sense of fun because it optimizes their fitness as hunters. But kittens are adaptation executors, evolution is the fitness optimizer. It's a subtle but important distinction. See http://www.overcomingbias.com/2007/11/adaptation-exec.html Terren They're adaptation executors, not fitness optimizers. --- On Mon, 8/25/08, Matt Mahoney [EMAIL PROTECTED] wrote: Kittens play with small moving objects because it teaches them to be better hunters. Play is not a goal in itself, but a subgoal that may or may not be a useful part of a successful AGI design. -- Matt Mahoney, [EMAIL PROTECTED] - Original Message From: Mike Tintner [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Monday, August 25, 2008 8:59:06 AM Subject: Re: [agi] How Would You Design a Play Machine? Brad, That's sad. The suggestion is for a mental exercise, not a full-scale project. And play is fundamental to the human mind-and-body - it characterises our more mental as well as more physical activities - drawing, designing, scripting, humming and singing scat in the bath, dreaming/daydreaming much more. It is generally acknowledged by psychologists to be an essential dimension of creativity - which is the goal of AGI. It is also an essential dimension of animal behaviour and animal evolution. Many of the smartest companies have their play areas. But I'm not aware of any program or computer design for play - as distinct from elaborating systematically and methodically or genetically on themes - are you? In which case it would be good to think about one - it'll open your mind give you new perspectives. This should be a group where people are not too frightened to play around with ideas. Brad: Mike Tintner wrote: ...how would you design a play machine - a machine that can play around as a child does? I wouldn't. IMHO that's just another waste of time and effort (unless it's being done purely for research purposes). It's a diversion of intellectual and financial resources that those serious about building an AGI any time in this century cannot afford. I firmly believe if we had not set ourselves the goal of developing human-style intelligence (embodied or not) fifty years ago, we would already have a working, non-embodied AGI. Turing was wrong (or at least he was wrongly interpreted). Those who extended his imitation test to humanoid, embodied AI were even more wrong. We *do not need embodiment* to be able to build a powerful AGI that can be of immense utility to humanity while also surpassing human intelligence in many ways. To be sure, we want that AGI to be empathetic with human intelligence, but we do not need to make it equivalent (i.e., just like us). I don't want to give the impression that a non-Turing intelligence will be easy to design and build. It will probably require at least another twenty years of two steps forward, one step back effort. So, if we are going to develop a non-human-like, non-embodied AGI within the first quarter of this century, we are going to have to just say no to Turing and start to use human intelligence as an inspiration, not a destination. Cheers, Brad Mike Tintner wrote: Just a v. rough, first thought. An essential requirement of an AGI is surely that it must be able to play - so how would you design a play machine - a machine that can play around as a child does? You can rewrite the brief as you choose, but my first thoughts are - it should be able to play with a) bricks b)plasticine c) handkerchiefs/ shawls d) toys [whose function it doesn't know] and e) draw. Something that should be soon obvious is that a robot will be vastly more flexible than a computer, but if you want to do it all on computer, fine. How will it play - manipulate things every which way? What will be the criteria of learning - of having done something interesting? How do infants, IOW, play? --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
On Mon, Aug 25, 2008 at 9:22 PM, Terren Suydam [EMAIL PROTECTED] wrote: Actually, kittens play because it's fun. Evolution has equipped them with the rewarding sense of fun because it optimizes their fitness as hunters. But kittens are adaptation executors, evolution is the fitness optimizer. It's a subtle but important distinction. See http://www.overcomingbias.com/2007/11/adaptation-exec.html Saying that play is not adaptive requires some backing (I expect it plays some role, so you need to be more convincing). -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.com/ --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Mike, I agree with Brad somewhat, because I do not think copying human (or animal) intellect is the goal. It is a means to the end of general intelligence. However, that certainly doesn't stop me from participating in a thought experiment. I think the big thing with artificial play is figuring out a good goal-creation scheme. My definition of play directly follows from this intuition: play is activity that results from a system that is rapidly changing its goals. In other words, play is behavior that is goal-oriented, but barely. The definition should probably be somewhat more specific-- when playing, people and animals don't just adopt totally arbitrary goals; we seem to prefer interesting goals. This is because there is a hidden biological agenda-- learning. But, learning is not *our* goal. Out goal is whatever arbitrary goal we have adopted for the purpose of play. One system I know of does something like this-- the PURR-PUSS system. Its rule is simple: if an unexpected event happens once, then the system will adopt the goal of trying to get it to happen again, by recreating the circumstances that led to it the first time. In carrying out the attempt, it should be able to greatly refine its concept of the circumstances that led to it-- because many of its attempts to recreate the event will probably fail. Many of these curiosity-based goals may be active at once. Since this is the system's only motivational factor, it could be called an artificial playing system (at least by my definition). -Abram On Mon, Aug 25, 2008 at 8:59 AM, Mike Tintner [EMAIL PROTECTED] wrote: Brad, That's sad. The suggestion is for a mental exercise, not a full-scale project. And play is fundamental to the human mind-and-body - it characterises our more mental as well as more physical activities - drawing, designing, scripting, humming and singing scat in the bath, dreaming/daydreaming much more. It is generally acknowledged by psychologists to be an essential dimension of creativity - which is the goal of AGI. It is also an essential dimension of animal behaviour and animal evolution. Many of the smartest companies have their play areas. But I'm not aware of any program or computer design for play - as distinct from elaborating systematically and methodically or genetically on themes - are you? In which case it would be good to think about one - it'll open your mind give you new perspectives. This should be a group where people are not too frightened to play around with ideas. Brad: Mike Tintner wrote: ...how would you design a play machine - a machine that can play around as a child does? I wouldn't. IMHO that's just another waste of time and effort (unless it's being done purely for research purposes). It's a diversion of intellectual and financial resources that those serious about building an AGI any time in this century cannot afford. I firmly believe if we had not set ourselves the goal of developing human-style intelligence (embodied or not) fifty years ago, we would already have a working, non-embodied AGI. Turing was wrong (or at least he was wrongly interpreted). Those who extended his imitation test to humanoid, embodied AI were even more wrong. We *do not need embodiment* to be able to build a powerful AGI that can be of immense utility to humanity while also surpassing human intelligence in many ways. To be sure, we want that AGI to be empathetic with human intelligence, but we do not need to make it equivalent (i.e., just like us). I don't want to give the impression that a non-Turing intelligence will be easy to design and build. It will probably require at least another twenty years of two steps forward, one step back effort. So, if we are going to develop a non-human-like, non-embodied AGI within the first quarter of this century, we are going to have to just say no to Turing and start to use human intelligence as an inspiration, not a destination. Cheers, Brad Mike Tintner wrote: Just a v. rough, first thought. An essential requirement of an AGI is surely that it must be able to play - so how would you design a play machine - a machine that can play around as a child does? You can rewrite the brief as you choose, but my first thoughts are - it should be able to play with a) bricks b)plasticine c) handkerchiefs/ shawls d) toys [whose function it doesn't know] and e) draw. Something that should be soon obvious is that a robot will be vastly more flexible than a computer, but if you want to do it all on computer, fine. How will it play - manipulate things every which way? What will be the criteria of learning - of having done something interesting? How do infants, IOW, play? --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by
Re: [agi] The Necessity of Embodiment
Hi Vlad, Thanks for taking the time to read my article and pose excellent questions. My attempts at answers below. --- On Sun, 8/24/08, Vladimir Nesov [EMAIL PROTECTED] wrote: On Sun, Aug 24, 2008 at 5:51 PM, Terren Suydam What is the point of building general intelligence if all it does is takes the future from us and wastes it on whatever happens to act as its goal? Indeed. Personally, I have no desire to build anything smarter than humans. That's a deal with the devil, so to speak, and one I believe most ordinary folks would be afraid to endorse, especially if they were made aware of the risks. The Singularity is not an inevitability, if we demand approaches that are safe in principle. And self-modifying approaches are not safe, assuming that they could work. I do however revel in the possibility of creating something that we must admit is intelligent in a general sense. Achieving such a goal would go a long way towards understanding our own abilities. So for me it's about research and understanding, with applications towards improving the quality of life. I advocate the slow and steady evolutionary approach because we can control the process (if not the agent) at each step of the way. We can stop the process at any point, study it, and make decisions about when and how to proceed. I'm all for limiting the intelligence of our creations before they ever get to the point that they can build their own or modify themselves. I'm against self-modifying approaches, largely because I don't believe it's possible to constrain their actions in the way Eliezer hopes. Iterative, recursive processes are generally emergent and unpredictable (the interesting ones, anyway). Not sure what kind of guarantees you could make for such systems in light of such emergent unpredictability. The problem with powerful AIs is that they could get their goals wrong and never get us the chance to fix that. And thus one of the fundamental problems that Friendliness theory needs to solve is giving us a second chance, building in deep down in the AI process the dynamic that will make it change itself to be what it was supposed to be. All the specific choices and accidental outcomes need to descend from the initial conditions, be insensitive to what went horribly wrong. This ability might be an end in itself, the whole point of building an AI, when considered as applying to the dynamics of the world as a whole and not just AI aspect of it. After all, we may make mistakes or be swayed by unlucky happenstance in all matters, not just in a particular self-vacuous matter of building AI. I don't deny the possibility of disaster. But my stance is, if the only approach you have to mitigate disaster is being able to control the AI itself, well, the game is over before you even start it. It seems profoundly naive to me that anyone could, even in principle, guarantee a super-intelligent AI to renormalize, in whatever sense that means. Then you have the difference between theory and practice... just forget it. Why would anyone want to gamble on that? Right, in a way that suggests you didn't grasp what I was saying, and that may be a failure on my part. That's why I was exploring -- I didn't get what you meant, and I hypothesized a coherent concept that seemed to fit what you said. I still don't understand that concept. Maybe I'll try again some other time if I can increase my own clarity on the concept. http://machineslikeus.com/news/design-bad-or-why-artificial-intelligence-needs-artificial-life (answering to the article) Creating *an* intelligence might be good in itself, but not good enough and too likely with negative side effects like wiping out the humanity to sum out positive in the end. It is a tasty cookie with black death in it. With the evolutionary approach, there is no self-modification. The agent never has access to its own code, because it's a simulation, not a program. So you don't have these hard take-off scenarios. However, it is very slow and that isn't appealing. AI folks want intelligence and they want it now. If the Singularity occurs to the detriment of the human race, it will be because of this rush to be the first to build something intelligent. I take some comfort in my belief that quick approaches simply won't succeed, but I admit I'm not 100% confident in that belief. You can't assert that we are not closer to AI than 50 years ago -- it's just unclear how closer we are. Great many techniques were developed in these years, and some good lessons learned the wrong way. Is it useful? Most certainly some of it, but how can we tell... Fair enough. It's a minor point though. Intelligence was created by a blind idiot evolutionary process that has no foresight and no intelligence. Of course it can be designed. Intelligence is all that evolution is, but immensely faster, better and flexible. In certain domains,
Re: [agi] How Would You Design a Play Machine?
Play is a form a strategy testing in an environment that doesn't severely penalize failures. As such, every AGI will necessarily spend a lot of time playing. If you have some other particular definition, then perhaps I could understand your response if you were to define the term. OTOH, if this is interpreted as being a machine that doesn't do anything BUT play (using my supplied definition), then your response has some merit, but even that can be very useful. Almost all of mathematics, e.g., is derived out of such play. I have a strong suspicion that machines that don't have a play mode can never proceed past the reptilian level of mentation. (Here I'm talking about thought processes that are mediated via the reptile brain in entities like mammals. Actual reptiles may have some more advanced faculties of which I'm unaware. (Note that, e.g., shrews don't have much play capability, but they have SOME.) Brad Paulsen wrote: Mike Tintner wrote: ...how would you design a play machine - a machine that can play around as a child does? I wouldn't. IMHO that's just another waste of time and effort (unless it's being done purely for research purposes). It's a diversion of intellectual and financial resources that those serious about building an AGI any time in this century cannot afford. I firmly believe if we had not set ourselves the goal of developing human-style intelligence (embodied or not) fifty years ago, we would already have a working, non-embodied AGI. Turing was wrong (or at least he was wrongly interpreted). Those who extended his imitation test to humanoid, embodied AI were even more wrong. We *do not need embodiment* to be able to build a powerful AGI that can be of immense utility to humanity while also surpassing human intelligence in many ways. To be sure, we want that AGI to be empathetic with human intelligence, but we do not need to make it equivalent (i.e., just like us). I don't want to give the impression that a non-Turing intelligence will be easy to design and build. It will probably require at least another twenty years of two steps forward, one step back effort. So, if we are going to develop a non-human-like, non-embodied AGI within the first quarter of this century, we are going to have to just say no to Turing and start to use human intelligence as an inspiration, not a destination. Cheers, Brad Mike Tintner wrote: Just a v. rough, first thought. An essential requirement of an AGI is surely that it must be able to play - so how would you design a play machine - a machine that can play around as a child does? You can rewrite the brief as you choose, but my first thoughts are - it should be able to play with a) bricks b)plasticine c) handkerchiefs/ shawls d) toys [whose function it doesn't know] and e) draw. Something that should be soon obvious is that a robot will be vastly more flexible than a computer, but if you want to do it all on computer, fine. How will it play - manipulate things every which way? What will be the criteria of learning - of having done something interesting? How do infants, IOW, play? --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: Information theoretic approaches to AGI (was Re: [agi] The Necessity of Embodiment)
Matt, What is your opinion on Goedel machines? http://www.idsia.ch/~juergen/goedelmachine.html --Abram On Sun, Aug 24, 2008 at 5:46 PM, Matt Mahoney [EMAIL PROTECTED] wrote: Eric Burton [EMAIL PROTECTED] wrote: These have profound impacts on AGI design. First, AIXI is (provably) not computable, which means there is no easy shortcut to AGI. Second, universal intelligence is not computable because it requires testing in an infinite number of environments. Since there is no other well accepted test of intelligence above human level, it casts doubt on the main premise of the singularity: that if humans can create agents with greater than human intelligence, then so can they. I don't know for sure that these statements logically follow from one another. They don't. I cannot prove that there is no non-evolutionary model of recursive self improvement (RSI). Nor can I prove that there is. But it is a question we need to answer before an evolutionary model becomes technically feasible, because an evolutionary model is definitely unfriendly. Higher intelligence bootstrapping itself has already been proven on Earth. Presumably it can happen in a simulation space as well, right? If you mean the evolution of humans, that is not an example of RSI. One requirement of friendly AI is that an AI cannot alter its human-designed goals. (Another is that we get the goals right, which is unsolved). However, in an evolutionary environment, the parents do not get to choose the goals of their children. Evolution chooses goals that maximize reproductive fitness, regardless of what you want. I have challenged this list as well as the singularity and SL4 lists to come up with an example of a mathematical, software, biological, or physical example of RSI, or at least a plausible argument that one could be created, and nobody has. To qualify, an agent has to modify itself or create a more intelligent copy of itself according to an intelligence test chosen by the original. The following are not examples of RSI: 1. Evolution of life, including humans. 2. Emergence of language, culture, writing, communication technology, and computers. 3. A chess playing (or tic-tac-toe, or factoring, or SAT solving) program that makes modified copies of itself by randomly flipping bits in a compressed representation of its source code, and playing its copies in death matches. 4. Selective breeding of children for those that get higher grades in school. 5. Genetic engineering of humans for larger brains. 1 fails because evolution is smarter than all of human civilization if you measure intelligence in bits of memory. A model of evolution uses 10^37 bits (10^10 bits of DNA per cell x 10^14 cells in the human body x 10^10 humans x 10^3 ratio of biomass to human mass). Human civilization has at most 10^25 bits (10^15 synapses in the human brain x 10^10 humans). 2 fails because individual humans are not getting smarter with each generation, at least not nearly as fast as civilization is advancing. Rather, there are more humans, and we are getting better organized through specialization of tasks. Human brains are not much different than they were 10,000 years ago. 3 fails because there are no known classes of problems that are provably hard to solve but easy to verify. Tic-tac-toe and chess have bounded complexity. It has not been proven that factoring is harder than multiplication. We don't know that P != NP, and even if we did, many NP-complete problems have special cases that are easy to solve, and we don't know how to program the parent to avoid these cases through successive generations. 4 fails because there is no evidence that above a certain level (about IQ 200) that childhood intelligence correlates with adult success. The problem is that adults of average intelligence can't agree on how success should be measured*. 5 fails for the same reason. *For example, the average person recognizes Einstein as a genius not because they are awed by his theories of general relativity, but because other people have said so. If you just read his papers (without understanding their great insights) and knew that he never learned to drive a car, you might conclude differently. -- Matt Mahoney, [EMAIL PROTECTED] --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
I'm not saying play isn't adaptive. I'm saying that kittens play not because they're optimizing their fitness, but because they're intrinsically motivated to (it feels good). The reason it feels good has nothing to do with the kitten, but with the evolutionary process that designed that adaption. It may seem like a minor distinction, but it helps to understand why, for example, people have sex with birth control. We don't have sex to maximize our genetic fitness, but because it feels good (or a thousand other reasons). We are adaption executers, not fitness optimizers. --- On Mon, 8/25/08, Vladimir Nesov [EMAIL PROTECTED] wrote: [EMAIL PROTECTED] wrote: Actually, kittens play because it's fun. Evolution has equipped them with the rewarding sense of fun because it optimizes their fitness as hunters. But kittens are adaptation executors, evolution is the fitness optimizer. It's a subtle but important distinction. See http://www.overcomingbias.com/2007/11/adaptation-exec.html Saying that play is not adaptive requires some backing (I expect it plays some role, so you need to be more convincing). -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.com/ --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Terren, Your broad distinctions are fine, but I feel you are not emphasizing the area of most interest for AGI, which is *how* we adapt rather than why. Interestingly, your blog uses the example of a screwdriver - Kauffman uses the same in Chap 12 of Reinventing the Sacred as an example of human creativity/divergence - i.e. our capacity to find infinite uses for a screwdriver. Do we think we could write an algorithm, an effective procedure, to generate a possibly infinite list of all possible uses of screwdrivers in all possible circumstances, some of which do not yet exist? I don't think we could get started. What emerges here, v. usefully, is that the capacity for play overlaps with classically-defined, and a shade more rigorous and targeted, divergent thinking, e.g. find as many uses as you can for a screwdriver, rubber teat, needle etc. ...How would you design a divergent (as well as play) machine that can deal with the above open-ended problems? (Again surely essential for an AGI) With full general intelligence, the problem more typically starts with the function-to-be-fulfilled - e.g. how do you open this paint can? - and only then do you search for a novel tool, like a screwdriver or another can lid. Terren: Actually, kittens play because it's fun. Evolution has equipped them with the rewarding sense of fun because it optimizes their fitness as hunters. But kittens are adaptation executors, evolution is the fitness optimizer. It's a subtle but important distinction. See http://www.overcomingbias.com/2007/11/adaptation-exec.html Terren They're adaptation executors, not fitness optimizers. --- On Mon, 8/25/08, Matt Mahoney [EMAIL PROTECTED] wrote: Kittens play with small moving objects because it teaches them to be better hunters. Play is not a goal in itself, but a subgoal that may or may not be a useful part of a successful AGI design. -- Matt Mahoney, [EMAIL PROTECTED] - Original Message From: Mike Tintner [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Monday, August 25, 2008 8:59:06 AM Subject: Re: [agi] How Would You Design a Play Machine? Brad, That's sad. The suggestion is for a mental exercise, not a full-scale project. And play is fundamental to the human mind-and-body - it characterises our more mental as well as more physical activities - drawing, designing, scripting, humming and singing scat in the bath, dreaming/daydreaming much more. It is generally acknowledged by psychologists to be an essential dimension of creativity - which is the goal of AGI. It is also an essential dimension of animal behaviour and animal evolution. Many of the smartest companies have their play areas. But I'm not aware of any program or computer design for play - as distinct from elaborating systematically and methodically or genetically on themes - are you? In which case it would be good to think about one - it'll open your mind give you new perspectives. This should be a group where people are not too frightened to play around with ideas. Brad: Mike Tintner wrote: ...how would you design a play machine - a machine that can play around as a child does? I wouldn't. IMHO that's just another waste of time and effort (unless it's being done purely for research purposes). It's a diversion of intellectual and financial resources that those serious about building an AGI any time in this century cannot afford. I firmly believe if we had not set ourselves the goal of developing human-style intelligence (embodied or not) fifty years ago, we would already have a working, non-embodied AGI. Turing was wrong (or at least he was wrongly interpreted). Those who extended his imitation test to humanoid, embodied AI were even more wrong. We *do not need embodiment* to be able to build a powerful AGI that can be of immense utility to humanity while also surpassing human intelligence in many ways. To be sure, we want that AGI to be empathetic with human intelligence, but we do not need to make it equivalent (i.e., just like us). I don't want to give the impression that a non-Turing intelligence will be easy to design and build. It will probably require at least another twenty years of two steps forward, one step back effort. So, if we are going to develop a non-human-like, non-embodied AGI within the first quarter of this century, we are going to have to just say no to Turing and start to use human intelligence as an inspiration, not a destination. Cheers, Brad Mike Tintner wrote: Just a v. rough, first thought. An essential requirement of an AGI is surely that it must be able to play - so how would you design a play machine - a machine that can play around as a child does? You can rewrite the brief as you choose, but my first thoughts are - it should be able to play with a) bricks b)plasticine c) handkerchiefs/ shawls d) toys [whose function it doesn't know] and e) draw. Something that should be soon obvious is that a
Re: [agi] How Would You Design a Play Machine?
On Mon, Aug 25, 2008 at 11:17 PM, Terren Suydam [EMAIL PROTECTED] wrote: I'm not saying play isn't adaptive. I'm saying that kittens play not because they're optimizing their fitness, but because they're intrinsically motivated to (it feels good). The reason it feels good has nothing to do with the kitten, but with the evolutionary process that designed that adaption. It may seem like a minor distinction, but it helps to understand why, for example, people have sex with birth control. We don't have sex to maximize our genetic fitness, but because it feels good (or a thousand other reasons). We are adaption executers, not fitness optimizers. The word because was misplaced. Cats hunt mice because they were designed to, and they were designed to, because it's adaptive. Saying that a particular cat instance hunts because it feels good is not very explanatory, like saying that it hunts because such is its nature or because the laws of physics drive the cat physical configuration through the hunting dynamics. Evolutionary design, on the other hand, is the point of explanation for the complex adaptation, the simple regularity in the Nature that causally produced the phenomenon we are explaining. -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.com/ --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] The Necessity of Embodiment
2008/8/25 Terren Suydam [EMAIL PROTECTED]: --- On Sun, 8/24/08, Vladimir Nesov [EMAIL PROTECTED] wrote: On Sun, Aug 24, 2008 at 5:51 PM, Terren Suydam wrong. This ability might be an end in itself, the whole point of building an AI, when considered as applying to the dynamics of the world as a whole and not just AI aspect of it. After all, we may make mistakes or be swayed by unlucky happenstance in all matters, not just in a particular self-vacuous matter of building AI. I don't deny the possibility of disaster. But my stance is, if the only approach you have to mitigate disaster is being able to control the AI itself, well, the game is over before you even start it. It seems profoundly naive to me that anyone could, even in principle, guarantee a super-intelligent AI to renormalize, in whatever sense that means. Then you have the difference between theory and practice... just forget it. Why would anyone want to gamble on that? You may be interested in goedel machines. I think this roughly fits the template that Eliezer is looking for, something that reliably self modifies to be better. http://www.idsia.ch/~juergen/goedelmachine.html Although he doesn't like explicit utility functions, the provably better is something he want. Although what you would accept as axioms for the proofs upon which humanity fate rests I really don't know. Personally I think strong self-modification is not going to be useful, the very act of trying to understand the way the code for an intelligence is assembled will change the way that some of that code is assembled. That is I think that intelligences have to be weakly self modifying, in the same way bits of the brain rewire themselves locally and subconciously, so to, AI will need to have the same sort of changes in order to keep up with humans. Computers at the moment can do lots of things better that humans (logic, bayesian stats), but are really lousy at adapting and managing themselves so the blind spots of infallible computers are always exploited by slow and error prone, but changeable, humans. Will Pearson --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Hi Mike, As may be obvious by now, I'm not that interested in designing cognition. I'm interested in designing simulations in which intelligent behavior emerges. But the way you're using the word 'adapt', in a cognitive sense of playing with goals, is different from the way I was using 'adaptation', which is the result of an evolutionary process. Terren --- On Mon, 8/25/08, Mike Tintner [EMAIL PROTECTED] wrote: From: Mike Tintner [EMAIL PROTECTED] Subject: Re: [agi] How Would You Design a Play Machine? To: agi@v2.listbox.com Date: Monday, August 25, 2008, 3:41 PM Terren, Your broad distinctions are fine, but I feel you are not emphasizing the area of most interest for AGI, which is *how* we adapt rather than why. Interestingly, your blog uses the example of a screwdriver - Kauffman uses the same in Chap 12 of Reinventing the Sacred as an example of human creativity/divergence - i.e. our capacity to find infinite uses for a screwdriver. Do we think we could write an algorithm, an effective procedure, to generate a possibly infinite list of all possible uses of screwdrivers in all possible circumstances, some of which do not yet exist? I don't think we could get started. What emerges here, v. usefully, is that the capacity for play overlaps with classically-defined, and a shade more rigorous and targeted, divergent thinking, e.g. find as many uses as you can for a screwdriver, rubber teat, needle etc. ...How would you design a divergent (as well as play) machine that can deal with the above open-ended problems? (Again surely essential for an AGI) With full general intelligence, the problem more typically starts with the function-to-be-fulfilled - e.g. how do you open this paint can? - and only then do you search for a novel tool, like a screwdriver or another can lid. Terren: Actually, kittens play because it's fun. Evolution has equipped them with the rewarding sense of fun because it optimizes their fitness as hunters. But kittens are adaptation executors, evolution is the fitness optimizer. It's a subtle but important distinction. See http://www.overcomingbias.com/2007/11/adaptation-exec.html Terren They're adaptation executors, not fitness optimizers. --- On Mon, 8/25/08, Matt Mahoney [EMAIL PROTECTED] wrote: Kittens play with small moving objects because it teaches them to be better hunters. Play is not a goal in itself, but a subgoal that may or may not be a useful part of a successful AGI design. -- Matt Mahoney, [EMAIL PROTECTED] - Original Message From: Mike Tintner [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Monday, August 25, 2008 8:59:06 AM Subject: Re: [agi] How Would You Design a Play Machine? Brad, That's sad. The suggestion is for a mental exercise, not a full-scale project. And play is fundamental to the human mind-and-body - it characterises our more mental as well as more physical activities - drawing, designing, scripting, humming and singing scat in the bath, dreaming/daydreaming much more. It is generally acknowledged by psychologists to be an essential dimension of creativity - which is the goal of AGI. It is also an essential dimension of animal behaviour and animal evolution. Many of the smartest companies have their play areas. But I'm not aware of any program or computer design for play - as distinct from elaborating systematically and methodically or genetically on themes - are you? In which case it would be good to think about one - it'll open your mind give you new perspectives. This should be a group where people are not too frightened to play around with ideas. Brad: Mike Tintner wrote: ...how would you design a play machine - a machine that can play around as a child does? I wouldn't. IMHO that's just another waste of time and effort (unless it's being done purely for research purposes). It's a diversion of intellectual and financial resources that those serious about building an AGI any time in this century cannot afford. I firmly believe if we had not set ourselves the goal of developing human-style intelligence (embodied or not) fifty years ago, we would already have a working, non-embodied AGI. Turing was wrong (or at least he was wrongly interpreted). Those who extended his imitation test to humanoid, embodied AI were even more wrong. We *do not need embodiment* to be able to build a powerful AGI that can be of immense utility to humanity while also surpassing human intelligence in many ways. To be sure, we want that AGI to be empathetic with human intelligence, but we do not need to make it equivalent (i.e., just like us). I don't want to give the impression that a non-Turing intelligence will be easy to design and build. It will
Re: [agi] How Would You Design a Play Machine?
Saying that a particular cat instance hunts because it feels good is not very explanatory Even if I granted that, saying that a particular cat plays to increase its hunting skills is incorrect. It's an important distinction because by analogy we must talk about particular AGI instances. When we talk about, for instance, whether an AGI will play, will it play because it's trying to optimize its fitness, or because it is motivated in some other way? We have to be that precise if we're talking about design. Terren --- On Mon, 8/25/08, Vladimir Nesov [EMAIL PROTECTED] wrote: The word because was misplaced. Cats hunt mice because they were designed to, and they were designed to, because it's adaptive. Saying that a particular cat instance hunts because it feels good is not very explanatory, like saying that it hunts because such is its nature or because the laws of physics drive the cat physical configuration through the hunting dynamics. Evolutionary design, on the other hand, is the point of explanation for the complex adaptation, the simple regularity in the Nature that causally produced the phenomenon we are explaining. -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.com/ --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
On Mon, Aug 25, 2008 at 12:52 PM, Vladimir Nesov [EMAIL PROTECTED] wrote: The word because was misplaced. Cats hunt mice because they were designed to, and they were designed to, because it's adaptive. And the adaption they have evolved in to, uses a pleasure process as a motivator. Saying that a particular cat instance hunts because it feels good is not very explanatory, like saying that it hunts because such is its nature or because the laws of physics drive the cat physical configuration through the hunting dynamics. Not at all. It defines the process by drives a cat to hunt, and also to practice - ie - play hunting. This is opposed to hunting due to reflex, like, say, a venus flytrap. I am reminded of a possibly apocryphal story about picasso: -- A woman asks Picasso to draw something for her on a napkin. He puts down a few lines, and says That will be $10,000. What! says the woman, That only took you five seconds to draw. No, that took me 40 years to draw. - Cats has evolved to see the process as a goal or reward in itself, over and above the requirements for food: If cats just hunted because they were hungry, they would never spend their downtime during kittenhood practicing and watching other cats hunt, and wouldn't be any good at hunting. And the result has many more advantages than simply optimising it's hunting strategies: it has evolved a cat that bonds with it's fellow kittens, learns to cooperate, and ultimately, becomes a better hunter because it sees the process of hunting as a game. Jonathan El-Bizri --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
On Tue, Aug 26, 2008 at 12:19 AM, Terren Suydam [EMAIL PROTECTED] wrote: Saying that a particular cat instance hunts because it feels good is not very explanatory Even if I granted that, saying that a particular cat plays to increase its hunting skills is incorrect. It's an important distinction because by analogy we must talk about particular AGI instances. When we talk about, for instance, whether an AGI will play, will it play because it's trying to optimize its fitness, or because it is motivated in some other way? We have to be that precise if we're talking about design. Of course. Different optimization processes at work, different causes. Let's say (ignoring if it's actually so for the sake of illustration) that cat plays because it provides it with developmental advantage through training its nervous system, giving it better hunting skills, and so an adaptation that drives cat to play was chosen *by evolution*. Cat doesn't play because *it* reasons that it would give it superior hunting skills, cat plays because of the emotional drive installed by evolution (or a more general drive inherent in its cognitive dynamics). When AGI plays to get better at some skill, it may be either a result of programmer's advice, in which case play happens because *programmer* says so, or as a result of its own conclusion that play helps with skills, and if skills are desirable, play inherits the desirability. In the last case, play happens because AGI decides so, which in turn happens because there is a causal link from play to a desirable state of having superior skills. -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.com/ --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Terren:As may be obvious by now, I'm not that interested in designing cognition. I'm interested in designing simulations in which intelligent behavior emerges.But the way you're using the word 'adapt', in a cognitive sense of playing with goals, is different from the way I was using 'adaptation', which is the result of an evolutionary process. Two questions: 1) how do you propose that your simulations will avoid the kind of criticisms you've been making of other systems of being too guided by programmers' intentions? How can you set up a simulation without making massive, possibly false assumptions about the nature of evolution? 2) Have you thought about the evolution of play in animals? (We play BTW with just about every dimension of activities - goals, rules, tools, actions, movements.. ). --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] The Necessity of Embodiment
Hi Will, I don't doubt that provable-friendliness is possible within limited, well-defined domains that can be explicitly defined and hard-coded. I know chess programs will never try to kill me. I don't believe however that you can prove friendliness within a framework that has the robustness required to make sense of a dynamic, unstable world. The basic problem, as I see it, is that Friendliness is a moving target, and context dependent. It cannot be defined within the kind of rigorous logical frameworks required to prove such a concept. Terren --- On Mon, 8/25/08, William Pearson [EMAIL PROTECTED] wrote: You may be interested in goedel machines. I think this roughly fits the template that Eliezer is looking for, something that reliably self modifies to be better. http://www.idsia.ch/~juergen/goedelmachine.html Although he doesn't like explicit utility functions, the provably better is something he want. Although what you would accept as axioms for the proofs upon which humanity fate rests I really don't know. Personally I think strong self-modification is not going to be useful, the very act of trying to understand the way the code for an intelligence is assembled will change the way that some of that code is assembled. That is I think that intelligences have to be weakly self modifying, in the same way bits of the brain rewire themselves locally and subconciously, so to, AI will need to have the same sort of changes in order to keep up with humans. Computers at the moment can do lots of things better that humans (logic, bayesian stats), but are really lousy at adapting and managing themselves so the blind spots of infallible computers are always exploited by slow and error prone, but changeable, humans. Will Pearson --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
If an AGI played because it recognized that it would improve its skills in some domain, then I wouldn't call that play, I'd call it practice. Those are overlapping but distinct concepts. Play, as distinct from pactice, is its own reward - the reward felt by a kitten. The spirit of Mike's question, I think, was about identifying the essential goalless-ness of play, the sense in which playing fosters adaptivity of goals. If you really want to interpret goal-satisfaction in play, it must be a meta-goal of mastering one's environment - and that is such a broadly defined goal that I don't see how one could specify it to a seed AI. I believe that's why evolution used the trick of making it fun. Terren --- On Mon, 8/25/08, Vladimir Nesov [EMAIL PROTECTED] wrote: Of course. Different optimization processes at work, different causes. Let's say (ignoring if it's actually so for the sake of illustration) that cat plays because it provides it with developmental advantage through training its nervous system, giving it better hunting skills, and so an adaptation that drives cat to play was chosen *by evolution*. Cat doesn't play because *it* reasons that it would give it superior hunting skills, cat plays because of the emotional drive installed by evolution (or a more general drive inherent in its cognitive dynamics). When AGI plays to get better at some skill, it may be either a result of programmer's advice, in which case play happens because *programmer* says so, or as a result of its own conclusion that play helps with skills, and if skills are desirable, play inherits the desirability. In the last case, play happens because AGI decides so, which in turn happens because there is a causal link from play to a desirable state of having superior skills. -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.com/ --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Hi Mike, Comments below... --- On Mon, 8/25/08, Mike Tintner [EMAIL PROTECTED] wrote: Two questions: 1) how do you propose that your simulations will avoid the kind of criticisms you've been making of other systems of being too guided by programmers' intentions? How can you set up a simulation without making massive, possibly false assumptions about the nature of evolution? Because I don't care about individual agents. Agents that fail to meet the requirements the environment demands, die. There's going to be a lot of death in my simulations. The risk I take is that nothing ever survives and I fail to demonstrate the feasibility of the approach. 2) Have you thought about the evolution of play in animals? (We play BTW with just about every dimension of activities - goals, rules, tools, actions, movements.. ). Not much. Play is such an advanced concept in intelligence, and my aims are far lower than that. I don't realistically expect to survive to see the evolution of human intelligence using the evolutionary approach I'm talking about. Terren --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
On Tue, Aug 26, 2008 at 1:26 AM, Terren Suydam [EMAIL PROTECTED] wrote: If an AGI played because it recognized that it would improve its skills in some domain, then I wouldn't call that play, I'd call it practice. Those are overlapping but distinct concepts. Play, as distinct from pactice, is its own reward - the reward felt by a kitten. The spirit of Mike's question, I think, was about identifying the essential goalless-ness of play, the sense in which playing fosters adaptivity of goals. If you really want to interpret goal-satisfaction in play, it must be a meta-goal of mastering one's environment - and that is such a broadly defined goal that I don't see how one could specify it to a seed AI. I believe that's why evolution used the trick of making it fun. What do you mean by trick? Fun of playing is evolutionary encoded, no tricks. You can try to encode it into a seed AI by adding a reference to an actual kitten in a right way, saying fun is that thing over there! without saying what it is explicitly, and providing this AI with a kitten. How to do it technically is of course a Friendly AI-complete problem, but its solution doesn't need to include all the fine points of the fun concept itself. On this subject, see: http://www.overcomingbias.com/2008/08/mirrors-and-pai.html -- in what sense AI can be as a mirror for complex concept instead of a pencil sketch explicitly hacked together by programmers; http://www.overcomingbias.com/2008/08/unnatural-categ.html -- why morality concept needs to be transferred in all details, and can't be learned from few examples; http://www.overcomingbias.com/2008/08/computations.html -- what a real-life concept may look like. -- Vladimir Nesov [EMAIL PROTECTED] http://causalityrelay.wordpress.com/ --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
On Mon, Aug 25, 2008 at 2:26 PM, Terren Suydam [EMAIL PROTECTED] wrote: If an AGI played because it recognized that it would improve its skills in some domain, then I wouldn't call that play, I'd call it practice. Those are overlapping but distinct concepts. The evolution of play is how nature has convinced us to practice skills of a general but un-predefinable type. Would it make sense to think of practice as the narrow AI version of play? Part of play is the specification of arbitrary goals and limitations within the overlying process. Games without rules aren't 'fun' to people or kittens. Play, as distinct from pactice, is its own reward - the reward felt by a kitten. The spirit of Mike's question, I think, was about identifying the essential goalless-ness of play, the sense in which playing fosters adaptivity of goals. If you really want to interpret goal-satisfaction in play, it must be a meta-goal of mastering one's environment - and that is such a broadly defined goal that I don't see how one could specify it to a seed AI. I believe that's why evolution used the trick of making it fun. But making it 'fun' doesn't answer the question of what the implicit goals are. Piaget's theories of assimilation can bring us closer to this, I am of the mind that they encompass at least part of the intellectual drive toward play and investigation. Jonathan El-Bizri --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Where is the hard dividing line between designed cognition and designed simulation (where intelligent behavior is intended to be emergent in both cases)? Even if an approach is taken where everything possible is done allow a 'natural' type evolution of behavior, the simulation design and parameters will still influence the outcome, sometimes in unknown and unknowable ways. Any amount of guidance in such a simulation (e.g. to help avoid so many of the useless eddies in a fully open-ended simulation) amounts to designed cognition. That being said, I'm particularly interested in the OCF being used as a platform for 'pure simulation' (Alife and more sophisticated game theoretical simulations), and finding ways to work the resulting experience and methods into the OCP design, which is itself a hybrid approach (designed cognition + designed simulation) intended to take advantage of the benefits of both. -dave On 8/26/08, Mike Tintner [EMAIL PROTECTED] wrote: Terren:As may be obvious by now, I'm not that interested in designing cognition. I'm interested in designing simulations in which intelligent behavior emerges.But the way you're using the word 'adapt', in a cognitive sense of playing with goals, is different from the way I was using 'adaptation', which is the result of an evolutionary process. Two questions: 1) how do you propose that your simulations will avoid the kind of criticisms you've been making of other systems of being too guided by programmers' intentions? How can you set up a simulation without making massive, possibly false assumptions about the nature of evolution? 2) Have you thought about the evolution of play in animals? (We play BTW with just about every dimension of activities - goals, rules, tools, actions, movements.. ). --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Terren: The spirit of Mike's question, I think, was about identifying the
essential goalless-ness of play..
Well, the key thing for me (although it was, technically, a play-ful
question :) ) is the distinction between programmed/planned exploration of
a basically known environment and ad hoc exploration of a deeply unknown
environment. In many ways, it follows on from my previous thread on
Philosophy of Learning in AGI, which asked - how do you learn an unfamiliar
subject/skill/ activity - could any definite set of principles guide you?
(This, I presume, is what Ben is somehow dealing with).
If you're an infant, or even often an adult, you don't know what this
strange object is for or how to manipulate it - so how do you go about
moving it and testing its properties? How do you go about moving your hand,
(or manipulator if you're a robot)? {I'd be interested in Bob M's input
here] - exploring its properties and capacities for movement too? What are
the principles if any that should constrain you?
Equally, if you're exploring an environment - a new kind of room, or a new
kind of territory like a garden, wood, forest, how do you go about moving
through it, deciding on paths, orienting yourself, mapping etc.? Remember
that these are initially alien environments, so the adult or AGI equivalent
is exploring a strange planet, or videogame world with alien kinds of laws.
Play - divergent thinking - exploration - these are all overlapping
dimensions of a general intelligence developing its intelligence, and
central to AGI.
And for the more abstractly inclined, I should point out that these
questions easily translate into the most abstract forms - like how do you
explore a new area of, or for, logic, or maths? How do you go about
exploring, or developing, a maths of, say, abstract art?
---
agi
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/
Modify Your Subscription:
https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51
Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Jonathan El-Bizri wrote: On Mon, Aug 25, 2008 at 2:26 PM, Terren Suydam [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote: If an AGI played because it recognized that it would improve its skills in some domain, then I wouldn't call that play, I'd call it practice. Those are overlapping but distinct concepts. The evolution of play is how nature has convinced us to practice skills of a general but un-predefinable type. Would it make sense to think of practice as the narrow AI version of play? No. Because in practice one is honing skills with a definite chosen purpose (and usually no instinctive guide), whereas in play one is honing skills without the knowledge that one is doing so. It's very different, e.g., to play a game of chess, and to practice playing chess. Part ... Jonathan El-Bizri --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] The Necessity of Embodiment
Is friendliness really so context-dependent? Do you have to be human to act friendly at the exception of acting busy, greedy, angry, etc? I think friendliness is a trait we project onto things pretty readily implying it's wired at some fundamental level. It comes from the social circuits, it's about being considerate or inocuous. But I don't know On 8/25/08, Terren Suydam [EMAIL PROTECTED] wrote: Hi Will, I don't doubt that provable-friendliness is possible within limited, well-defined domains that can be explicitly defined and hard-coded. I know chess programs will never try to kill me. I don't believe however that you can prove friendliness within a framework that has the robustness required to make sense of a dynamic, unstable world. The basic problem, as I see it, is that Friendliness is a moving target, and context dependent. It cannot be defined within the kind of rigorous logical frameworks required to prove such a concept. Terren --- On Mon, 8/25/08, William Pearson [EMAIL PROTECTED] wrote: You may be interested in goedel machines. I think this roughly fits the template that Eliezer is looking for, something that reliably self modifies to be better. http://www.idsia.ch/~juergen/goedelmachine.html Although he doesn't like explicit utility functions, the provably better is something he want. Although what you would accept as axioms for the proofs upon which humanity fate rests I really don't know. Personally I think strong self-modification is not going to be useful, the very act of trying to understand the way the code for an intelligence is assembled will change the way that some of that code is assembled. That is I think that intelligences have to be weakly self modifying, in the same way bits of the brain rewire themselves locally and subconciously, so to, AI will need to have the same sort of changes in order to keep up with humans. Computers at the moment can do lots of things better that humans (logic, bayesian stats), but are really lousy at adapting and managing themselves so the blind spots of infallible computers are always exploited by slow and error prone, but changeable, humans. Will Pearson --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] The Necessity of Embodiment
Eric, We're talking Friendliness (capital F), a convention suggested by Eliezer Yudkowsky, that signifies the sense in which an AI does no harm to humans. Yes, it's context dependent. Do no harm is the mantra within the medical community, but clearly there are circumstances in which you do a little harm to achieve greater health in the long run. Chemotherapy is a perfect example. Would we trust an AI if it proposed something like chemotherapy? Before we understood that to be a valid treatment, would we really believe it was being Friendly? You want me to drink *what*? Or take any number of ethical dilemmas, in which it's ok to steal food if it's to feed your kids. Or killing ten people to save twenty. etc. How do you define Friendliness in these circumstances? Depends on the context. Terren --- On Mon, 8/25/08, Eric Burton [EMAIL PROTECTED] wrote: Is friendliness really so context-dependent? Do you have to be human to act friendly at the exception of acting busy, greedy, angry, etc? I think friendliness is a trait we project onto things pretty readily implying it's wired at some fundamental level. It comes from the social circuits, it's about being considerate or inocuous. But I don't know On 8/25/08, Terren Suydam [EMAIL PROTECTED] wrote: Hi Will, I don't doubt that provable-friendliness is possible within limited, well-defined domains that can be explicitly defined and hard-coded. I know chess programs will never try to kill me. I don't believe however that you can prove friendliness within a framework that has the robustness required to make sense of a dynamic, unstable world. The basic problem, as I see it, is that Friendliness is a moving target, and context dependent. It cannot be defined within the kind of rigorous logical frameworks required to prove such a concept. Terren --- On Mon, 8/25/08, William Pearson [EMAIL PROTECTED] wrote: You may be interested in goedel machines. I think this roughly fits the template that Eliezer is looking for, something that reliably self modifies to be better. http://www.idsia.ch/~juergen/goedelmachine.html Although he doesn't like explicit utility functions, the provably better is something he want. Although what you would accept as axioms for the proofs upon which humanity fate rests I really don't know. Personally I think strong self-modification is not going to be useful, the very act of trying to understand the way the code for an intelligence is assembled will change the way that some of that code is assembled. That is I think that intelligences have to be weakly self modifying, in the same way bits of the brain rewire themselves locally and subconciously, so to, AI will need to have the same sort of changes in order to keep up with humans. Computers at the moment can do lots of things better that humans (logic, bayesian stats), but are really lousy at adapting and managing themselves so the blind spots of infallible computers are always exploited by slow and error prone, but changeable, humans. Will Pearson --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Hi Johnathon, I disagree, play without rules can certainly be fun. Running just to run, jumping just to jump. Play doesn't have to be a game, per se. It's simply a purposeless expression of the joy of being alive. It turns out of course that play is helpful for achieving certain goals that we interpret as being installed by evolution. But we don't play to achieve goals, we do it because it's fun. As Mike said, this very discussion is a kind of play, and while we can certainly identify goals that we try to accomplish in the course of hashing these things out, there's an element in it, for me anyway, of just doing it because I love doing it. I suspect that's true for others here. I hope so, anyway. Of course, those that are dogmatically functionalist will view such language as 'fun' as totally irrelevant. That's ok. The cool thing about AI is that eventually, it will shed light on whether subjective experience (to functionalists, an inconvenience to be done away with) is critical to intelligence. To address your second question, the implicit goal is always reproduction. If there is one basic reductionist element to all of life, it is that. Making play fun is a way of getting us to play at all, so that we are more likely to reproduce. There's a limit however to the usefulness and accuracy of reducing everything to reproduction. Terren --- On Mon, 8/25/08, Jonathan El-Bizri [EMAIL PROTECTED] wrote: Part of play is the specification of arbitrary goals and limitations within the overlying process. Games without rules aren't 'fun' to people or kittens. Play, as distinct from pactice, is its own reward - the reward felt by a kitten. The spirit of Mike's question, I think, was about identifying the essential goalless-ness of play, the sense in which playing fosters adaptivity of goals. If you really want to interpret goal-satisfaction in play, it must be a meta-goal of mastering one's environment - and that is such a broadly defined goal that I don't see how one could specify it to a seed AI. I believe that's why evolution used the trick of making it fun. But making it 'fun' doesn't answer the question of what the implicit goals are. Piaget's theories of assimilation can bring us closer to this, I am of the mind that they encompass at least part of the intellectual drive toward play and investigation. Jonathan El-Bizri agi | Archives | Modify Your Subscription --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
Re: [agi] How Would You Design a Play Machine?
Hi David, Any amount of guidance in such a simulation (e.g. to help avoid so many of the useless eddies in a fully open-ended simulation) amounts to designed cognition. No, it amounts to guided evolution. The difference between a designed simulation and a designed cognition is the focus on the agent itself. In the latter, you design the agent and turn it loose, testing it to see if it does what you want it to. In the former (the simulation), you turn a bunch of candidate agents loose and let them compete to do what you want them to. The ones that don't, die. You're specifying the environment, not the agent. If you do it right, you don't even have to specify the goals. With designed cognition, you must specify the goals, either directly (un-embodied), or in some meta-fashion (embodied). Terren --- On Mon, 8/25/08, David Hart [EMAIL PROTECTED] wrote: From: David Hart [EMAIL PROTECTED] Subject: Re: [agi] How Would You Design a Play Machine? To: agi@v2.listbox.com Date: Monday, August 25, 2008, 6:04 PM Where is the hard dividing line between designed cognition and designed simulation (where intelligent behavior is intended to be emergent in both cases)? Even if an approach is taken where everything possible is done allow a 'natural' type evolution of behavior, the simulation design and parameters will still influence the outcome, sometimes in unknown and unknowable ways. Any amount of guidance in such a simulation (e.g. to help avoid so many of the useless eddies in a fully open-ended simulation) amounts to designed cognition. That being said, I'm particularly interested in the OCF being used as a platform for 'pure simulation' (Alife and more sophisticated game theoretical simulations), and finding ways to work the resulting experience and methods into the OCP design, which is itself a hybrid approach (designed cognition + designed simulation) intended to take advantage of the benefits of both. -dave On 8/26/08, Mike Tintner [EMAIL PROTECTED] wrote: Terren:As may be obvious by now, I'm not that interested in designing cognition. I'm interested in designing simulations in which intelligent behavior emerges.But the way you're using the word 'adapt', in a cognitive sense of playing with goals, is different from the way I was using 'adaptation', which is the result of an evolutionary process. Two questions: 1) how do you propose that your simulations will avoid the kind of criticisms you've been making of other systems of being too guided by programmers' intentions? How can you set up a simulation without making massive, possibly false assumptions about the nature of evolution? 2) Have you thought about the evolution of play in animals? (We play BTW with just about every dimension of activities - goals, rules, tools, actions, movements.. ). --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?; Powered by Listbox: http://www.listbox.com agi | Archives | Modify Your Subscription --- agi Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/ Modify Your Subscription: https://www.listbox.com/member/?member_id=8660244id_secret=111637683-c8fa51 Powered by Listbox: http://www.listbox.com
