Date: Thu 2 Jun 1988 01:40-EDT From: AIList Moderator Nick Papadakis Reply-To: AIList@AI.AI.MIT.EDU Us-Mail: MIT Mail Stop 38-390, Cambridge MA 02139 Phone: (617) 253-2737 Subject: AIList Digest V7 #14 To: AIList@AI.AI.MIT.EDU Status: R AIList Digest Thursday, 2 Jun 1988 Volume 7 : Issue 14 Today's Topics: Still More Free Will ---------------------------------------------------------------------- Date: 30 May 88 14:41:18 GMT From: mcvax!ruuinf!piet@uunet.uu.net (Piet van Oostrum) Subject: Re: More Free Will In article <532@wsccs.UUCP> dharvey@wsccs.UUCP (David Harvey) writes: If free will has created civilization as we know it, then it must be accepted with mixed emotions. This means that Hitler, Stalin, some of the Catholic Popes during the middle ages and others have created a great deal of havoc that was not good. One of the prime reasons for AI is to perhaps develop systems that prevent things like this from happening. If we with our free will (you said it, not me) can't seem to create a decent world to live in, perhaps a machine without free will operating within prescribed boundaries may do a better job. We sure haven't done too well. I agree we haven't done too well, but if these same persons (i.e. WE) are going to design a machine, what make you think this machine will do a better job??? If the machine doesn't have a free will, the designers must decide what kind of decisions it will make, and it will be based upon their insights, ideas, moral etc. Or would you believe AI researchers (or scientists in general) are inherently better than rulers, popes, nazi's, communists or catholics, to name a few? Hitler and Stalin had scientists work for them, and there are now AI researchers working on war-robots and similar nasty things. That doesn't give ME much hope from that area. -- Piet van Oostrum, Dept of Computer Science, University of Utrecht Padualaan 14, P.O. Box 80.089, 3508 TB Utrecht, The Netherlands Telephone: +31-30-531806 UUCP: ...!mcvax!ruuinf!piet ------------------------------ Date: 30 May 88 16:40:28 GMT From: dvm@yale-zoo.arpa (Drew Mcdermott) Subject: Free will More on the self-modeling theory of free will: Since no one seems to have understood my position on this topic, I will run the risk that no one cares about my position, and try to clarify. Sometimes parties to this discussion talk as if "free will" were a new kind of force in nature. (As when Biep Durieux proposed that free will might explain probability rather than vice versa.) I am sure I misrepresent the position; the word "force" is surely wrong here (as is the word "new"). The misrepresentation is unavoidable; this kind of dualism is simply not a live option for me. Nor can I see why it needs to be a perenially live option on an AI discussion bulletin board. So, as I suggested earlier, let's focus on the question of free will within the framework of Artificial Intelligence. And here it seems to me the question is, How would we tell an agent with free will from an agent without it? Two major strands of the discussion seem completely irrelevant from this standpoint: (1) Determinism vs. randomness. The world is almost certainly not deterministic, according to quantum mechanics. Quantum mechanics may be false, but Newtonian mechanics is certainly false, so the evidence that the world is deterministic is negligible. (Unless the Everett-Wheeler interpretation of quantum mechanics is true, in which case the world is a really bizarre place.) So, if determinism is all that's bothering you, you can relax. Actually, I think what's really bothering people is the possibility of knowledge (traditionally, divine knowledge) of the outcomes of their future decisions, which has nothing to do with determinism. (2) My introspections about my ability to control my thoughts or whatnot. There is no point in basing the discussion on such evidence, until we have a theory of what conscious thoughts are. Such a theory must itself start from the outside, looking at a computational agent in the world and explaining what it means for it to have conscious thoughts. That's a fascinating topic, but I think we can solve the free will problem with less trouble. So, what makes a system free? To the primitive mind, free decisions are ubiquitous. A tornado decides to blow my house down; it is worth trying to influence its decision with various rewards or threats. But nowadays we know that the concept of decision is just out of place in reasoning about tornados. The proper concepts are causal; if we can identify enough relevant antecedent factors, we can predict (and perhaps someday control) the tornado's actions. Quantum mechanics and chaos set limits to how finely we can predict, but that is irrelevant. Now we turn to people. Here it seems as if there is no need to do away with the idea of decision, since people are surely the paradigmatic deciders. But perhaps that attitude is "unscientific." Perhaps the behaviorists are right, and the way we think about thunderstorms is the right way to think about people. If that's the actual truth, then we should be tough-minded and acknowledge it. It is *not* the truth. Freedom gets its toehold from the fact that it is impossible for an agent to think of itself in terms of causality. Contrast my original bomb scenario with this one: R sees C wander into the blast area, and go up to the bomb. R knows that C knows all about bombs, and R knows that C has plenty of time to save itself, so R decides to do nothing. (Assume that preventing the destruction of other robots gets big points in R's utility function.) In this case, R is reasoning about an agent other than itself. Its problem is to deduce what C will actually do, and what C will actually suffer. The conclusion is that C will prosper, so R need do nothing. It would be completely inappropriate for R to reason this way about itself. Suppose R comes to realize that it is standing next to a bomb, and it reasons as follows: R knows all about bombs, and has plenty of time to save itself, so I need do nothing. Its reasoning is fallacious, because it is of the wrong kind. R is not being called on to deduce what R will do, but to be a part of the causal fabric that determines what R will do, in other words: to make a decision. It is certainly possible for a robot to engage in a reasoning pattern of this faulty kind, but only by pretending to make a decision, inferring that the decision will be made like that, and then not carrying it out (and thus making the conclusion of the inference false). Of course, such a process is not that unusual; it is called "weakness of the will" by philosophers. But it is not the sort of thing one would be tempted to call an actual decision. An actual decision is a process of comparative evaluation of alternatives, in a context where the outcome of the comparison will actually govern behavior. (A robot cannot decide to stop falling off a cliff, and an alcoholic or compulsive may not actually make decisions about whether to cease his self-destructive behavior.) This scenario is one way for a robot to get causality wrong when reasoning about itself, but there is a more fundamental way, and that is to just not notice that R is a decision maker at all. With this misperception, R could tally its sources of knowledge about all influences on R's behavior, but it would miss the most important one, namely, the ongoing alternative-evaluation process. Of course, there are circumstances in which this process is in fact not important. If R is bound and gagged and floating down a river, then it might as well meditate on hydrodynamics, and not work on a decision. But most of the time the decision-making process of the robot is actually one of the causal antecedents of its future. And hence, to repeat the central idea, *there is no point in trying to think causally about oneself while making a decision that is actually part of the causal chain. Any system that realizes this has free will.* This theory accounts for why an agent must think of itself as outside the causal order of things when making a decision. However, it need not think of other agents this way. An agent can perfectly well think of other agents' behavior as caused or uncaused to the same degree the behavior of a thunderstorm is caused or uncaused. There is a difference: One of the best ways to cause a decision-making agent to do something is to give him a good reason to do it, whereas this strategy won't work with thunderstorms. Hence, an agent will do well to sort other systems into two categories, those that make free decisions and those that don't, and deal with them differently. By the way, once a decision is made there is no problem with its maker thinking of it purely causally, in exactly the same way it thinks about other decision makers. An agent can in principle see *all* of the causal factors going into its own past decisions, although in practice the events of the past will be too random or obscure for an exhaustive analysis. It is surely not dehumanizing to be able to bemoan that if only such-and-such had been brought to my attention, I would have decided otherwise than I did, but, since it wasn't, I was led inexorably to a wrong decision. Now let me deal with various objections: (1) Some people said I had neglected the ability of computers to do reflexive meta-reasoning. As usual, the mention of meta-reasoning makes my head swim, but I shall try to respond. Meta-reasoning can mean almost anything, but it usually means escaping from some confining deductive system in order to reason about what that system ought to conclude. If this is valuable, there is no reason not to use it. But my picture is of a robot faced with the possibility of reasoning about itself as a physical system, which is in general a bad thing to do. The purpose of causal-exemption flagging is to shut pointless reasoning down, meta or otherwise. So, when O'Keefe says: So the mere possibility of an agent having to appear to simulate itself simulating itself ... doesn't show that unbounded resources would be required: we need to know more about the nature of the model and the simulation process to show that. I am at a loss. Any system can simulate itself with no trouble. It could go over past or future decisions with a fine-tooth comb, if it wanted to. What's pointless is trying to simulate the present period of time. Is an argument needed here? Draw a mental picture: The robot starts to simulate, and finds itself simulating ... the start of a simulation. What on earth could it mean for a system to figure out what it's doing by simulating itself? (2) Free will seems on this theory to have little to do with consciousness or values. Indeed it does not. I think a system could be free and not be conscious at all; and it could certainly be free and not be moral. What is the minimal level of free will? Consider a system for scheduling the movement of goods into and out of a warehouse. It has to synchronize its shipments with those of other agents, and let us suppose that it is given those other shipments in the form of various schedules that it must just work around. From its point of view, the shipments of other agents are caused, and its own shipments are to be selected. Such a system has what we might call *rudimentary* free will. To get full-blown free will, we have to suppose that the system is able to notice the discrepancy between boxes that are scheduled to be moved by someone else, and boxes whose movements depend on its decisions. I can imagine all sorts of levels of sophistication in understanding (or misunderstanding) the discrepancy, but just noticing it is sufficient for a system to have full-blown free will. At that point, it will have to realize that it and its tools (the things it moves in the warehouse) are exempt from causal modeling. (3) Andrew Gelsey has pointed out that a system might decide what to do by means other than simulating various alternative courses of action. For instance, a robot might decide how hard to hit a billiard ball by solving an equation for the force required. In this case, the asymmetry appears in what is counted as an independent variable (i.e., the force administered). And if the robot notices and appreciates the asymmetry, it is free. (4) David Sher has objected If I understand [McDermott's theory] correctly it runs like this: To plan one has a world model including future events. Since you are an element of the world then you must be in the model. Since the model is a model of future events then your future actions are in the model. This renders planning unnecessary. Thus your own actions must be excised from the model for planning to avoid this "singularity." Taken naively, this analysis would prohibit multilevel analyses such as is common in game theory. A chess player could not say things like if he moves a6 then I will move Nc4 or Bd5 which will lead .... The response to this misreading should be obvious. There are two ways to think about my future actions. One way is to treat them as conditional actions, begun now, and not really future actions at all. (Cf. the notion of strategy in game theory.) The more interesting insight is that an agent can reason about its future actions as if they were those of another agent. There is no problem with doing this; the future is much like the past in this respect, except we have less information about it. A robot could reason at its leisure about what decision it would probably make if confronted with some future situation, and it could use an arbitrarily detailed simulation of itself to do this reasoning, provided it has time to run it before the decision is to be made. But all of this self-prediction is independent of actually making the decision. When the time comes to actually make it, the robot will find itself free again. It will not be bound by the results of its simulation. This may seem like a nonsequitur; how could a robot not faithfully execute its program the same way each time it is run? There is no need to invoke randomness; the difference between the two runs is that the second one is in a context where the results of the simulation are available. Of course, there are lots of situations where the decision would be made the same way both times, but all we require is that the second be correctly classified as a real -- free -- decision. I find Sher's "fix" to my theory more dismaying: However we can still make the argument that Drew was making its just more subtle than the naive analysis indicates. The way the argument runs is this: Our world model is by its very nature a simplification of the real world (the real world doesn't fit in our heads). Thus our world model makes imperfect predictions about the future and about consequence. Our self model inside our world model shares in this imperfection. Thus our self model makes inaccurate predictions about our reactions to events. We perceive ourselves as having free will when our self model makes a wrong prediction. This is not at all what I meant, and seems pretty shaky on its own merits. This theory makes an arbitrary distinction between an agent's mistaken predictions about itself and its mistaken predictions about other systems. I think it's actually a theory of why we tend to attribute free will to so many systems, including thunderstorms. We know our freedom makes us hard to predict, and so we attribute freedom to any system we make a wrong prediction about. This kind of paranoia is probably healthy until proven false. But the theory doesn't explain what we think free will is in the first place, or what its explanatory force is in explaining wrong predictions. Free will is not due to ignorance. Imagine that the decision maker is a robot with a very routine environment, so that it often has complete knowledge both of its own listing and of the external sensory data it will be receiving prior to a decision. So it can simulate itself to any level of detail, and it might actually do that, thinking about decisions in advance as a way of saving time later when the actual decision had to be made. None of this would allow it to avoid making free decisions. -- Drew McDermott ------------------------------ Date: Mon, 30 May 88 15:05:47 EDT From: Bharat.Dave@CAD.CS.CMU.EDU Subject: reconciling free will and determinism Perhaps following quote from "Chaos" by J. Gleick may help reconcile apparently dichotomous concepts of determinism and free will. This is more of a *metaphor* rather than a rigorous argument. Discussing the work of a group of people at UCSC on chaotic systems, the author quotes Doyne Farmer[pg.251]. Farmer said, "On a philosophical level, it struck me as an operational way to define free will, in a way that allowed you to reconcile free will with determinism. The system is deterministic, but you can't say what it's going to do next... Here was one coin with two sides. Here was order, with randomness emerging, and then one step further away was randomness with its own underlying order." Two other themes that pop up quite often in this book are apparent differences between behavior at micro and macro scales, and sensitive dependence on initial conditions on the behavior of dynamical systems. Choice of the vantage point significantly affects what you see. At molecular level, we are all a stinking mess with a highly boring and uniform behavior; on a different level, I can't predict with certainty how most people will respond to this message. Seems like you can't be free unless you acknowledge determinism :-) ------------------------------ Date: 31 May 88 03:24:17 GMT From: glacier!jbn@labrea.stanford.edu (John B. Nagle) Subject: Re: Free will Since this discussion has lost all relevance to anything anybody is likely to actually implement in the AI field in the next twenty years or so, could this be moved to talk.philosophy? John Nagle ------------------------------ Date: 31 May 88 14:10:08 GMT From: bwk@mitre-bedford.arpa (Barry W. Kort) Subject: Self Simulation Drew McDermott's lengthy posting included a curious nugget. Drew paints a scenario in which a robot engages in a simulation which includes the robot itself as a causal agent in the simulation. Drew asks, "What on earth could it mean for a system to figure out what it's doing by simulating itself?" I was captured by the notion of self-simulation, and started day-dreaming, imagining myself as an actor inside a simulation. I found that, as the director of the day-dream, I had to delegate free will to my simulated self. The movie free-runs, sans script. It was just like being asleep. So, perhaps a robot who engages in self-simulation is merely dreaming about itself. That's not so hard. I do it all the time. --Barry Kort ------------------------------ Date: 31 May 88 14:35:33 GMT From: uhccux!lee@humu.nosc.mil (Greg Lee) Subject: Re: Free will Edward Lasker, in his autobiographical book about his experiences as a chess master, describes a theory and philosophical tract by his famous namesake, Emmanuel Lasker, who was world chess champion for many years. It concerned a hypothetical being, the Macha"ide, which is so advanced and profound in its thought that its choices have become completely constrained. It can discern and reject all courses of action that are not optimal, and therefore it must. It is so evolved that it has lost free will. Greg, lee@uhccux.uhcc.hawaii.edu ------------------------------ Date: 31 May 88 14:39:41 GMT From: uvaarpa!virginia!uvacs!cfh6r@umd5.umd.edu (Carl F. Huber) Subject: Re: Free Will & Self Awareness In article <5323@xanth.cs.odu.edu> Warren E. Taylor writes: >In article <1176@cadre.dsl.PITTSBURGH.EDU>, Gordon E. Banks writes: > >"Spanking" IS, I repeat, IS a form of redesigning the behavior of a child. >Many children listen to you only when they are feeling pain or are anticipating >the feeling of pain if they do not listen. > Also, pain is often the only teacher a child will listen to. He >From what basis do you make this extraordinary claim? Experience? or do you have some reputable publications to mention - I would like to see the studies. I also assume that "some" refers to the 'average' child - not pathological exceptions. > I have an extremely hard-headed nephew >who "deserves" a spanking quite often because he is doing something that is >dangerous or cruel or simply socially unacceptable. He is also usually >maddeningly defiant. ^^^^^^^^^^^^^^^^^^^ Most two to six year olds are. How old is this child? 17? What other methods have been tried? Spanking most generally results from frustrated parents who beleive they have "tried everything", while they actually haven't begun to scratch the surface of what works. >learns to associate a certain action with undesirable consequences. I am not >the least bit religious, but the old Biblical saying of "spare the rod..." is right, so let's not start something we'll regret, ala .psychology. > >Flame away. > Warren. voila. cheers. -carl ------------------------------ Date: 31 May 88 15:27:11 GMT From: mailrus!caen.engin.umich.edu!brian@ohio-state.arpa (Brian Holtz) Subject: Re: DVM's request for definitions In article <1020@cresswell.quintus.UUCP>, ok@quintus.UUCP (Richard A. O'Keefe) writes: > In article <894@maize.engin.umich.edu>, brian@caen.engin.umich.edu (Brian > Holtz) writes: > > 3. volition: the ability to identify significant sets of options and to > > predict one's future choices among them, in the absence of any evidence > > that any other agent is able to predict those choices. > > > > There are a lot of implications to replacing free will with my notion of > > volition, but I will just mention three. > > > > - If my operationalization is a truly transparent one, then it is easy > > to see that volition (and now-dethroned free will) is incompatible with > > an omniscient god. Also, anyone who could not predict his behavior as > > well as someone else could predict it would no longer be considered to > > have volition. [proceeding excerpts not in any particular order] > For me, "to have free will" means something like "to act in accord with > my own nature". If I'm a garrulous twit, people will be able to predict > pretty confidently that I'll act like a garrulous twit (even though I > may not realise this), but since I will then be behaving as I wish I > will correctly claim free will. Recall that my definition of free will ("the ability to make at least some choices that are neither uncaused nor completely determined by physical forces") left little room for it to exist. Your definition (though I doubt you will appreciate being held to it this strictly) leaves too much room: doesn't a falling rock, or the average computer program, "act in accord with [its] own nature"? > One thing I thought AI people were taught was "beware of the homunculus". > As soon as we start identifying parts of our mental activity as external > to "ourselves" we're getting into homunculus territory. I agree that homunculi are to be avoided; that is why I relegated "the ability to make at least some choices that are neither uncaused nor completely determined by *external* physical forces" to being a definition not of free will, but of "self-determination". The free will that you are angling for sounds a lot like what I call self-determination, and I would welcome any efforts to sharpen the definition so as to avoid the externality/internality trap. So until someone comes up with a definition of free will that is better than yours and mine, I think the best course is to define free will out of existence and take my "volition" as the operationalized designated hitter for free will in our ethics. > What has free will to do with prediction? Presumably a dog is not > self conscious or engaged in predicting its activities, but does that > mean that a dog cannot have free will? Free will has nothing to do with prediction; volition does. The question of whether a dog has free will is a simple one with either your definition *or* mine. By my definition, nothing has free will; by yours, it seems to me that everything does. (Again, feel free to refine your definition if I've misconstrued it.) A dog would seem to have self-determination as I've defined it, but you and I agree that my definition's reliance on ex/in-ternality makes it a suspect categorization. A dog would clearly not have volition, since it can't make predictions about itself. And since volition is what I propose as the predicate we should use in ethics, we are happily exempt from extending ethical personhood to dogs. > "no longer considered to have volition ..." I've just been reading a > book called "Predictable Pairing" (sorry, I've forgotten the author's) > name, and if he's right it seems to me that a great many people do > not have volition in this sense. If we met Hoyle's "Black Cloud", and > it with its enormous computational capacity were to predict our actions > better than we did, would that mean that we didn't have volition any > longer, or that we had never had it? A very good question. It would mean that we no longer had volition, but that we had had it before. My notion of volition is contingent, because it depends on "the absence of any evidence that any other agent is able to predict" our choices. What is attractive to me about volition is that it would be very useful in answering ethical questions about the "free will" (in the generic ethical sense) of arbitrary candidates for personhood: if your AI system could demonstrate volition as defined, then your system would have met one of the necessary conditions for personhood. What is unnerving to me about my notion of volition is how contingent it is: if Hoyle's "Black Cloud" or some prescient god could foresee my behavior better than I could, I would reluctantly conclude that I do not even have an operational semblence of free will. My conclusion would be familiar to anyone who asserts (as I do) that the religious doctrine of predestination is inconsistent with believing in free will. I won't lose any sleep over this, though; Hoyle's "Black Cloud" would most likely need to use analytical techniques so invasive as to leave little of me left to rue my loss of volition. ------------------------------ Date: 31 May 88 15:51:48 GMT From: mind!thought!ghh@princeton.edu (Gilbert Harman) Subject: Re: Free will In article <17470@glacier.STANFORD.EDU> jbn@glacier.UUCP (John B. Nagle) writes: > > Since this discussion has lost all relevance to anything anybody >is likely to actually implement in the AI field in the next twenty years >or so, could this be moved to talk.philosophy? > > John Nagle Drew McDermott's suggestion seems highly relevant to implementations while offering a nice approach to at least one problem of free will. (It seems clear that people have been worried about a number of different things under the name of "free will".) How about keeping a discussion of McDermott's approach here and moving the rest of the discussion to talk.philosophy? Gilbert Harman Princeton University Cognitive Science Laboratory 221 Nassau Street, Princeton, NJ 08542 ghh@princeton.edu HARMAN@PUCC.BITNET ------------------------------ Date: Tue, 31 May 88 17:08 EDT From: GODDEN%gmr.com@RELAY.CS.NET Subject: RE: Free Will and Self-Awareness >From: bwk@mitre-bedford.arpa (Barry W. Kort) [...] >It is not clear to me whether aggression is instinctive (wired-in) >behavior or learned behavior. You might want to take a look at the book >On Aggression< by Konrad Lorenz. I don't have the complete reference with me, but can supply upon request. I read it back in 1979, but if I recall one of the primary theses set forth in the book is that agression is indeed instinctive in all animal life forms including humans and serves both to defend and perpetuate the species. Voluminous citations of purposeful and useful aggressive behavior in many species are provided. I think he also philosophizes on how we as thinking and peace-loving people (with free will (!)) can make use of our conscious recognition of our innate aggression to keep it at appropriate levels of manifestation. I became very excited about his ideas at the time. -Kurt Godden GM Research godden@gmr.com ------------------------------ Date: Tue, 31 May 88 18:48:56 -0700 From: peck@Sun.COM Subject: free will First of all, i what to say that i'm probably in this camp: > 4. (Minsky et al.) There is no such thing as free will. We can dispense > with the concept, but for various emotional reasons we would rather not. I haven't followed all this discussion, but it seems the free-will protagonists basic claim is that "choices" are made randomly or by a homonculus, and the physicists claim that the homonculus has no free will either. Level 1 (psychological): an intelligent agent first assumes that it has choices, and reasons about the results of each and determines predictively the "best" choice. ["best" determined by some optimization criterion function] This certainly looks like free-will, especially to those experiencing it or introspecting on it. Level 0 (physics): the process or computation which produces the reasoning and prediction at level 1 is deterministic. (plus or minus quantum effects) >From level 0 looking up to level 1, it's hard to see where free will comes in. Do the free-will'ists contend that the mind retroactively controls the laws of physics? More perspicuous would be that the mind distorts its own perception of itself to believe that it is immune to the laws of physics. My real questions is: Does it matter whether man or machine has free will? In what way does the existance of free will make a more intelligent actor, a better information processor, or a better controller of processes? If an agent makes good decisions or choices, or produces control outputs, stores or interprets information, or otherwise produces behaviors, What does it matter (external to that agent) whether it has free will? What does it matter *internal* to that agent? Does it matter if the agent *believes* it has free will? [For those systems sophisticated enough to introspect, *believing* to have free will is useful, at least initially (was this Minsky's argument?).] Are there any objective criteria for determining if an agent has free-will? [If, as i suspect, this whole argument is based on the assumption that: Free-will iff (not AI), then it seems more feasible to work on the AI vs (not AI) clause, and *then* debate the free-will clause.] My brief response to the moral implication of free will: For those agents that can accept the non-existance of free will and still function reliably and intelligently (ie the "enlightened few"), the concept is not necessary. For others, (the "majority") the concept of free will, like that of God, and sin, and laws, and purpose, and the rest of what Vedanta/Budhists would call the "Illusion", is either necessary or useful. In this case, it is important *not* to persuade someone that they do not have free will. If they can not figure out the reason and ramifications of that little gem of knowledge, it is probably better that they not be burdened. yes, a little knowledge *is* a dangerous thing. - --- My other bias: the purpose of intelligence to produce more flexible, more capable controllers of processes (the first obvious one is the biochemical plant that is your body). These controllers of course are mostly characterized by their control of information: how well they understand or model the environment or the controlled process determines how well they can predict or control that environment or process. So, intelligence is inexorably tied up with information processing (interpreting, storing, encoding/decodeing), and decision making. From this point of view, the more interesting questions are: What is the criterion function (utility function, feedback funtion, or whatever you call it); how is it formed; how is it modified; how is it evaluated? The question of evaluation is an important difference between artificial intelligence and organic intelligence. In people the evaluation is done by the "hardware", it is modified as the body is modified, bio-chemically. It is the same biochemistry that the evaluation function is trying to control! Thought for the day: If it turns out that the criterion function is designed to self-perpetuate *itself* (the function, not merely its agent), by arranging to have itself be modified as the results of the actions based on its predictive evaluations, [ie, it is self serving and self perpetuating just as genes/memes/virusus are] would that help explain why choices seem indeterminate and "free"? ------------------------------ Date: 1 Jun 88 09:38:00 EDT From: "CUGINI, JOHN" Reply-to: "CUGINI, JOHN" Subject: Free will - the metaphysics continues... Drew McDermott writes: > I believe most of the confusion about this concept comes from there > not being any agreed-upon "common sense" of the term "free will." To > the extent that there is a common consensus, it is probably in favor > of dualism, the belief that the absolute sway of physical law stops at > the cranium. Unfortunately, ever since the seventeenth century, the > suspicion has been growing among the well informed that this kind of > dualism is impossible. ... > > If we want to debate about AI versus dualism, ...we can. [Let's] > propose technical definitions of free will, or propose dispensing with > the concept altogether. ... > > I count four proposals on the table so far: > > 1. (Propose by various people) Free will has something to do with randomness. > > 2. (McCarthy and Hayes) When one says "Agent X can do action A," or > "X could have done A," > > 3. (McDermott) To say a system has free will is to say that it is > "reflexively extracausal," > > 4. (Minsky et al.) There is no such thing as free will. I wish to respond somewhat indirectly by trying to describe the "classical" position - ie to present a paradigm case of free will. Thus, we will have an account of *sufficient* conditions for free will. Readers may then consider whether these conditions are necessary - whether we can back off to a less demanding case, and still have free will. I suspect the correct answer is "no", but I'll not argue that point too thoroughly. *** *** *** *** *** *** Brief version: Free will is the ability of a conscious entity to make free decisions. The decision is free, in that, although the entity causes the decision, nothing causes the entity to make the decision. *** *** *** *** *** *** (Rough analogy: an alpha particle is emitted (caused) by the decay of a nucleus, but nothing caused the nucleus to decay and emit the particle - the emitting [deciding] is uncaused). There's an unfortunate ambiguity in the word "decision" - it can mean the outcome (the decision was to keep on going), or the process (his decision was swift and sure). Keeping these straight, it is the decision-process, the making of the decision-outcome, that is uncaused. The decision-outcome is, of course, caused (only) by the decision-process. Discussion: I'm going to opt for breadth at the expense of depth - please read in non-nitpicking spirit. 1. Randomness - free will is related to randomness only in that both are examples of acausality. True, my uncaused decision is "random" wrt to the physical world, and/or the history of my consciousness - ie not absolutely predictable therefrom. That doesn't mean it's random, in the stronger sense of being a meaningless event that popped out of nowhere - see next item. 2. Conscious entity - free will is a feature which only a conscious entity (like you and me, and maybe your dog) can have - can anyone credit an unconscious entity with free will, even if it "makes a random decision" in some derivative sense (eg a chess-playing program which uses QM processes to vary its moves) ? "Consciousness is a problematic concept" you say? - well, yes, but not as much so as is free will - and I think it's only problematic to those who insist that it be reduced to more simple concepts. There ain't any. Free decisions "pop out" of conscious entities, not nuclei. If you don't know what a conscious entity is, you're not reading this. No getting around it, free will brings us smack up to the problem of the self - if there are no conscious selves, there is no free will. While it may be difficult to describe what it takes to be a conscious self, at least we don't doubt the existence of selves, as we may the existence of free will. So the strategy here is to take selves as a given (for now at least) and then to say under what conditions these selves have free will. 3. Dualism - I believe we can avoid this debate; I maintain that free will requires consciousness. Whether consciousness is physical, we can leave aside. I can't resist noting that Saul Kripke is probably as "well-informed" as anyone, and last I heard, he was a dualist. It's quite fashionable nowadays to take easy verbal swipes at dualism as an emblem of one's sophistication. I suspect that some swipers might be surprised at how difficult it is to concoct a good argument against dualism. 4. Physics - Does the requirement for acausality require the violation of known physical laws? Very unclear. First, note that all kinds of macro-events are ultimately uncaused in that they stem from ontologically random quantum events (eg radiation causing birth defects, cancer...). Whether brain events magnify QM uncertainty in this way no one really knows, but it's not to be ruled out. Further, very little is understood of the causal relations between brain and consciousness (hence the dualism debate). At any rate, the position is that, for the conscious decision-process to be free, it must be uncaused. If this turns out to violate physical laws as presently understood, and if the present understanding turns out to be correct, then this just shows that there is no free will. 5. No denial of statistical probabilitities or influence - None of the above denies that allegedly free deciders are, in fact, quite predictable (probabilistically). It is highly unlikely that I will decide to put up my house for sale tomorrow, but I could. My conscious reasons for not doing so do not absolutely determine that I won't. I could choose to in spite of these reasons. 6. Free will as "could have decided otherwise" - This formulation is OK as long as the strength of the "could" includes at least physical possibility, not just logical possibility. If one could show that my physical brain-state today physically determines that I will (consciously) decide to sell my house tomorrow, it's not a free decision. 7. A feature, not an event - I guess most will agree that free will is a capability (like strong arms) which is manifested in particular events - in the case of free will, the events are free decisions. An entity might make some caused decisions and some free - free will only says he/she can make free ones, not that all his/her decisions are free. 8. Rationality / Intelligence - It may well be true that rationality makes free will worth having but there's no reason not to consider the will free even in the absence of a whole lot of intelligence. Rationality makes strong arms more valuable as well, but one can still have strong arms without it. As long as one can acausally decide, one has free will. 9. Finding out the truth - Need I mention that the above is intended to define what free will is, not necessarily to tell one how to go about determining whether it exists or not. To construct a test reflecting the above considerations is no small task. Moreover, one must decide (!) where the burden of proof lies: a) I feel free, therefore it's up to you to prove my feelings are illusory and that all my decision-processes are caused, or b) the "normal" scientific assumption is that all macro-events have proximate macro-causes, and therefore it's up to me to show that my conscious processes are a "special case" of some kind. John Cugini ------------------------------ Date: 1 Jun 88 14:39:00 GMT From: apollo!nelson_p%apollo.uucp@eddie.mit.edu Subject: Free will and self-awareness Gilbert Cockton posts: >The test is easy, look at the references. Do the same for AAAI and >IJCAI papers. The subject area seems pretty introspective to me. >If you looked at an Education conference proceedings, attended by people who >deal with human intelligence day in day out (rather than hack LISP), you >would find a wide range of references, not just specialist Education >references. >You will find a broad understanding of humanity, whereas in AI one can >often find none, just logical and mathematical references. I still >fail to see how this sort of intellectual background can ever be >regarded as adequate for the study of human reasoning. On what >grounds does AI ignore so many intellectual traditions? Because AI would like to make some progress (for a change!). I originally majored in psychology. With the exception of some areas in physiological pyschology, the field is not a science. Its models and definitions are simply not rigorous enough to be useful. This is understandable since the phenomena it attempts to address are far too complex for the currently available intellectual and technical tools. The result is that psychologists and sociologists waste much time and money over essentially unresolvable philosophical debates, sort of like this newsgroup! When you talk about an 'understanding of humanity' you clearly have a different use of the term 'understanding' in mind than I do. Let's move this topic to talk.philosophy!! --Peter Nelson ------------------------------ Date: Wed, 01 Jun 88 12:48:12 From: ALFONSEC%EMDCCI11.BITNET@CUNYVM.CUNY.EDU Subject: Free will et al. >Thanasis Kehagias >Subject: AI is about Death > . . . >SUGGESTED ANSWER: if AI is possible, then it is possible to create >intelligence. all it takes is the know-how and the hardware. also, the >following inference is not farfetched: intelligence -> life. so if AI >is possible, it is possible to give life to a piece of hardware. no ghost >in the machine. no soul. call this the FRANKENSTEIN HYPOTHESIS, or, for >short, the FH (it's just a name, folks!). The fact that we have "created intelligence" (i.e. new human beings) since thousands of years ago, has not stopped the current controversy or the discussion about the existence of the soul. If, sometime in the future, we make artificial intelligence beings, the discussions will go on the same as today. What is to prevent a machine from having a soul? The question cannot be decided in a discussion, because it comes from totally different axioms, or starting points. The (non-)existence of the soul (or of free will) is not a conclusion, but an axiom. It is much more difficult to convince people to change their axioms than to accept a fact. Regards, Manuel Alfonseca, ALFONSEC at EMDCCI11 ------------------------------ End of AIList Digest ********************