Can Machines be Persons?

Core 155

Spring 2005

This is a philosophical question, and as we noted in class, it is a question that we can use to interrogate our concept of person.  At the end of class, we got around to the “So what?” question, which is a very important in philosophy—perhaps the most important question—and I am not sure I addressed it adequately.  This handout is an attempt to do a better job.  Please feel free to let me know where this falls short, or how you disagree.  (The discussion group for the class would be a good place to do this, linked from the homepage.)

I.          Motivating the Question

A.                 Monstrous Motivation:  As I have argued over the last 24 weeks or so, monstrosity as a concept is rooted in our tendency to marginalize and even demonize that which we regard as other, i.e., as significantly and perhaps frighteningly different from ourselves.  If we hope to understand the “other”, we must gain an understanding of “self”, and this requires coming to understand what it is to be persons.  Thus, part of our education concerning the nature of monstrosity requires that we examine the concept person.

B.                 Why this question is important (class): The nature of personhood is at the heart of both Frankenstein and Blade Runner.  In the former, we are asked to consider whether intellect alone makes the man, whereas in the latter we are asked to consider (a) whether empathy is a necessary condition for personhood (i.e., can you be a person without it?), and (b) whether it makes sense to consider androids (i.e., machines) as persons if they are indistinguishable behaviorally from human beings.

C.                 Why this question is important (life):  The concept person is used extensively outside of the academy.  As I mentioned in class, “person” is a juridical term, which is to say that it is a term of law.  People debate whether fetuses are persons, for instance, with the full complement of moral rights; on the other side of life, we debate whether people in permanent vegetative states are persons.  When you accord something the status of person, there is the expectation that it has a certain legal and moral status, viz., the same moral and legal status that we human beings have.  It changes how we regard it and how we treat it. 

II.        Answering the Question

A.                 Method:  As we mentioned, this is a philosophical question.  What makes it different from “Can Idahoans be lottery winners?” is that it involves a concept that is in need of clarification, viz., person.  (One could question what it is to be a machine, but that would just make it more philosophical.) 

1.                  We clarify concepts like this by examining the things to which they apply (viz., the extension) and then abstracting away from these applications to the rule (viz., the intension) that governs application of the concept.  This process involves going back and forth from examination of the extension, to formulation of a candidate intension, to evaluation of whether the intension accounts for the extension. 

2.                  As this is conceptual analysis, the judge in the third stage is not observation (as it would be in science), but rather intuition.  We have to determine whether the rule we’ve identified does just to our intuitions about how the concept person is applied.  In the end, our work may force us to modify our intuition—say, if we can’t get our intuition to agree but there is ample independent evidence (perhaps from other people) that our rule must be correct.

B.                 Strategy:  We proceed by attempting to determine whether we can imagine circumstances in which a machine could be a person—i.e., circumstances in which we would treat a machine as having the same legal and moral standing that we do.  Would behavior suffice?  Would it need to be carbon-based and not silicon-based?  Would it need to be intellectual?  Emotional?  Would it need to demonstrate the ability to make moral decisions? 

1.                  We start by assuming that many machines aren’t persons and that most human beings are persons, if not all.  Think here of three Venn circles: one for machines, one for human beings, and one for persons.  These are three differentiable concepts: the first is a functional concept, the second a biological concept, and the third a juridical, religious, or perhaps social concept. We are out to determine the ways in which these circles relate, and in particular, if there is any overlap between the machine circle and the person circle.

2.                  Note that you might think all humans are persons and all persons are humans, but this doesn’t mean that the concept person is the same as the concept human being.  They might be co-extensional without being the same concept.  (Think here of equilateral triangle and equiangular triangle.)  Determining that they are co-extensional is a result­—it is something that analysis reveals and not something that can be assumed, given that they are different concepts.

3.                  Our answers here won’t suffice to specify fully the nature of the concept person, but it will enable us to determine whether there are boundary conditions in this vicinity beyond which the concept cannot extend. 

III.       Answers to the Question

A.                 Yes

1.                  In this case, the machine circle does overlap the person circle.  If you believe this, you are willing to consider the possibility that some day, we could legitimately accord the same legal and moral standing that we have to machines.  Note that this does not mean that you would actually have to do this—only that you consider it a possibility.

2.                  One way to come to this conclusion is if you think that certain machines could be human beings, perhaps because of their biological character, and all human beings are persons.  More likely, though, you will come to this conclusion because you think that not all persons are human beings, and that the overlap between person and machine occurs in the place where person and human being don’t intersect.

3.                  You might decide that only machines of a certain type could be persons, e.g., machines that are capable of having the same kind of experiences that we do, and perhaps machines that are forced to deal with their own mortality in the way that we are.

4.                  Examples of literary sources that suggest this picture are AI, Bicentennial Man, and (arguably) Blade Runner. 

B.                 No

1.                  Here you conclude that there is no overlap.  This conclusion will typically be driven by the belief that there is something about persons that machines cannot attain, i.e., a distinguishing characteristic.

2.                  Distinguishing Characteristics:

a.                   Biology—they can never have our biological character, which is in part based on our origins.

b.                  Intellect—they cannot reason as we do.

c.                   Emotion—they are incapable of an adequate level of emotional response.

d.                  Moral Sense—while they may calculate and deliberate, they will never have a true moral sense, perhaps because whether you have that or not depends on factors beyond your control.

e.                   Consciousness—machines will never be able to have the kinds of feelings and other conscious experiences that we can have.

f.                    Origin—what it is to be a person depends on how you come to be, and they cannot come to be in the appropriate way.

g.                   Others?

C.                 Notice that throughout this process, you must ask yourself whether you can imagine a machine that has this characteristic or not—is it possible?  If so, then you will not take the characteristic to differentiate machines and persons.  If you cannot find one, then you are left with the conclusion that machines could be persons.

IV.       Person-al Problems

A.                 Can’t you just say anything here?  No.  It isn’t the case that anything goes.  The process of conceptual analysis is a public one, and the concept person is part of the public domain. Granted, we do rely in analysis on intuition, but you correct for possible flaws in your own intuitions by comparing them with the intuitions of other people.  We are after knowledge here—we operate under the assumption that “person” is meaningful and that it has a meaning we can locate.  We do not know what that is at the beginning of the process, but after we engage in it, we hope to have it.

B.                 Isn’t the meaning of “person” relative?  Perhaps, but that is something we would come to after analysis.  It may be that this concept is used in different ways by different groups, and if so, then we should expect different answers to our question.  But without analysis, we won’t be able to tell for sure whether the concept is in fact relative.  (And keep in mind that disagreement does not imply relativity—it could imply that at most one of the parties to the disagreement is correct.)

C.                 What value does this have? Analysis of a concept as close to us as this is valuable in its own right—any knowledge we uncover here will feed into our worldview and inform us about the nature of the world.  But when a concept is used to support legislation, funding, information flow, and punishment—like the concept person—the stakes are much, much higher.  They are no longer merely epistemological.  Now they are moral, and in many cases, personal.  Analysis is required to make sure that mistakes aren’t made, given these stakes.