The Way To Dialogue

This paper was originally written and accepted for presentation at an MIT conference on Identity, Formation & Dignity as part of its project on God and Computers. Unfortunately I was unable to attend so the paper was not presented.

It is easy to see why theologians should be interested in Artificial Intelligence; it is less easy to see why those working in the field of Artificial Intelligence might be interested in theology. But if theology is serious about engaging AI and cognitive science in ongoing dialogue, the conversation must be more than one way. There are two strands of thought interwoven through this paper. The first discusses the need for theology to take a more substantial role in the dialogue process; the second sketches a few theological insights into personhood, illustrating ways in which theology should be contributing to AI.

The tremendous achievements made in the field of AI have sparked interest through numerous disciplines, including, as is evidenced by this conference, theology. As computer technology advances, machines are being created that are capable, among other things, of speech recognition, visual recognition, translating languages, and playing games – most notably chess, in which only last year the computer Deep Blue defeated world champion Gary Kasparov for the first time. Each of these advances contributes to the understanding of our physical bodies and, in particular, the workings of our brains. In turn, these influence our understanding of human identity.

The developments in cognitive science are already impacting on theology, raising ethical and hermeneutical questions; the influence of theology on cognitive science is not so evident. In the commercial world there is what is known as a WIIFM. (pr.wiffum) "What's In It For Me". If you approach a company for sponsorship, for example, you need to approach them with answers to their WIIFM if you hope to achieve your sponsorship goal. In Religious Studies, a key word is "reciprocity"; the essence is the same. Applying the basic principle of a WIIFM to our call for dialogue means that if we want AI and cognitive science to spend time and resources talking with us, we need to show them that theology is capable of making a valuable contribution to their tasks.

In the field of cognitive science, as the understanding of intelligence expands, debate continues over whether or not it is possible to achieve the ultimate in artificial intelligence – a conscious machine. In philosophical circles, and here in theology, the ongoing question of personhood is currently being addressed with renewed vigour. We want to know what it is that identifies us as human persons. And herein lies the WIIFM. Theology has the potential to play a significant role in the AI quest through its insights into the nature of personhood.

Historically, intelligent machines have not been an issue. Instead we humans have sought ways to distinguish ourselves from other animals, and theology has been a major contributor to such discussions. Across the years philosophers, scientists and theologians have told us, among other things, that "only humans have souls"; "only humans use tools"; "only humans can reason". One by one, however, each of these distinctions has been discredited. In this day and age, the concept of "soul" is given little credence for any animal - human or non-human. Animal behaviourists have shown us how some animals use tools, albeit in far less sophisticated ways than humans use them. Similarly, we are familiar with various reasoning capacities in different species. In each of these areas, humanity may be further to the right on a continuum, but there is no black and white divide as previously presumed, and we have learned to accept our place in the animal world, calling upon our "superior intellect" to justify our earthly dominance. And our intellect has served us well; we have linguistic skills far in advance of all other animals, and through the use of language, we are able to think in terms of abstracts and our problem-solving skills are second to none.

At least that was the case. As artificial intelligence continues to make headway, it is feasible that intelligent machines will one day be as far to the right of us on any intellectual continuum, as we are to cows. If we claim a right to earthly dominion on the basis of superior intellect, reason suggests that dominion should rightly pass to our intellectual successors - a prospect few of us would relish. And it is this prospect, I believe, that has revitalised our interest in personhood. The difference now, to our historical inquiries, is that we are keen, not to distinguish ourselves from other animals, but to discover (or create) an unassailable divide between human beings and intelligent non-biological machines.

Such a goal, of course, presupposes that some form of artificial mind will be achieved. This supposition gives rise to the question of how we will know if and when this ultimate goal of artificial intelligence has actually been reached. The Turing test is usually touted as the best available means to determine machine intelligence, but this test still leaves open the question of consciousness. We do not know what the relationship between intelligence and consciousness is.

Certainly, intelligence itself is no indication of consciousness or personhood. If, when we speak of intelligence, we refer to an ability to respond to environment, to learn from experience, solve problems and "think" laterally, in ways quantifiable by performance, then we are already successfully mass-producing artificially intelligent machines. I am convinced that with the appropriate software, even my old, out-moded 486 PC could out-perform me on any standard IQ test. I am even more convinced that my old PC is totally non-conscious, despite the many anthropomorphic behaviours I commonly attribute to it.

Intelligence might be considered a property of an active mind. It can also be thought of as a qualitative assessment of the performance of a mind. Robert Sternberg once made the observation that there are nearly as many definitions of intelligence as there are experts to define it. Intelligence, it would seem, is in the eye of the beholder. Furthermore, it seems unimportant that machines may be more 'intelligent' than we are if they are reliant on our agency. But it is the prospect of an artificial consciousness rather than artificial intelligence that stirs controversy, for it is the contemplation of a conscious machine that challenges our concept of personhood. And it is at this point that science needs to look beyond itself for an answer: If these intelligent machines develop consciousness and become intelligent agents themselves, at which point do we grant them the status of persons?

However we might understand the mind-body problem, cognitive science tells us that the mind is, in some way, a function of the brain. We can have a brain without a mind, but we cannot have a mind without a brain. Theologically, this sits well with the Hebrew concept of the human being as "an animated body". Indeed, from a theological perspective, human beings might be seen as the prototypical AI, formed, as we are, from inanimate and inorganic dust. Whatever else we might say about our personhood, theologically speaking, we are "created, intelligent, conscious beings".

One does not have to accept that humans (and other life forms) were divinely created to regard humankind in this way. Viewed simply through an evolutionary schema, we can still be considered models of "artificial intelligence". Indeed, Daniel Dennett says something similar when he claims that humans are conscious machines, made up of "natural robots" and descended from robots. The word "robots" in Dennett's context refers to unconscious life forms or biological automatons of the kinds that make up our bodies and from which we evolved. Dennett explains these automatons as biological entities that necessarily act in response to particular stimuli. Defined in this way, the term 'robot' appears to be a reasonable description for the cellular and organic components of our bodies. But what does it mean to say we are conscious, biological automatons? In other words, conscious, unconscious life forms? The temptation here is to cancel out the antonyms and simply say we are life forms, but to do this would be both unhelpful and, I think, a mistake. The interplay between consciousness and the unconscious mind would seem to be an essential element of what we call the self (whatever the 'self' may be).

If we take seriously the claim that human beings are conscious machines, (and I advocate we do), the task of theology would seem to run parallel to that of artificial intelligence. Whereas AI concentrates its efforts in understanding the workings of the mind, theology is a discipline which lays open our humanity and focuses on the human condition. Minds may be essential to personhood, but are they enough to ensure personhood? It is here that theology's insights might prove invaluable.

At this point I shall put forward just three theological insights that add to our understandings of personhood. No doubt, many of you could provide other examples, but my purpose here is not to present a full theological portrait of personhood; rather it is to indicate ways in which we might evoke the interest of science, and provide them with their WIIFM.

Firstly, theology emphasises the uniqueness of every person. As physical beings we can certainly be reproduced: we've been biologically reproducing our species since our origins, and it would appear that we now have the technical ability to artificially reproduce ourselves through cloning. But animal reproduction merely produces physical beings similar to the beings whose genetic information is reproduced. Personhood is not reproducible. Each newly produced being must develop its own unique personality. No matter how close the physical resemblance, theology insists on the uniqueness of each person.

Secondly, personhood is not something we are born with in the way that we are born with a brain or a liver. Although, no doubt our genetic make-up contributes greatly to the persons we become, theology deems personhood is evoked only in relationship. Genetics may well affect the relationships we build, but without relationships there can be no personhood. Of course, for the Christian or Jew, the primary relationship that calls all humans into personhood is the relationship we have with God, or rather, the relationship God has with us. And this includes the relationships we have with all persons, for we live our relationship with God through our relationships with others. In other words, my personhood has developed out of the personal relationships I share, and I am who I am because of those I am in relationship with. When one's life is touched by another, the person one is, is changed.

Thirdly, and perhaps most controversially, theology recognizes the fragmentary nature of the self. Most of us readily acknowledge human consciousness as fragmented and multi-faceted, but we tend to see our selves as continuous identities. The self of whom I am aware as I write these words appears, to me, to be clearly continuous with the self that sat down and began writing some time ago. My consciousness, however, has flitted between the words and ideas that make up this paper and a multitude of non-related thoughts from the temperature of my coffee, to the number of split ends in my hair, to distant friends, etc. Moreover, there were times that passed when I was not aware of being conscious of anything at all, yet it would be incorrect to say that I was unconscious through those periods. As thinking subjects, we have limited control over the contents of our consciousness. Some things can readily be called into conscious thought; some things will come with effort; other things, not at all. Equally, there are some things that cannot be wilfully eliminated from consciousness, and there are some times when we may be surprised by what springs unbidden into our consciousness. Nevertheless, it is only through one’s consciousness that one can know anything of the self. Consciousness then, may be considered an important element of the self, and something that emerges from the self, (John Puddefoot calls it the "thrown-ness of the self") but it cannot be equated with the self.

We look at selves – our own and others' – through our own fragmented windows of consciousness to see something akin to a series of video clips or chapters of a book strung together. When we watch a movie on television we see a series of film clips, each internally consistent, which, put together with skilful editing, tell us a coherent story, even if our viewing is interspersed with non-related episodes such as commercials. During the course of watching a film or reading a book, our attention is directed to different events in ways that indicate to us whether we are to understand them as simultaneous events, linear occurrences, or flashbacks, etc. Our minds somehow piece together these various fragments and fill in the spaces to give the illusion of continuity. We may feel as though we have followed the life of Moses from infancy to Exodus, kept company with Anne Boelyn through her reign of a thousand days, or struggled through a single day in the life of Ivan Denisovich, all within the space of a few hours. I suggest that our construction of self is a similar process. One's brain somehow selectively filters the various fragments of consciousness and edits them to construct coherent understandings of one’s own self and other selves. We could say a continuous self is epistemological rather than ontological.

So where is the reciprocity? Most disciplines, it seems, seek to provide all-embracing theories to produce a coherent picture of human nature. The underlying premise of such theories is that even the most irrational of human behaviours can be rationally explained. To this end, we can read theories which explain religious experience, for example, in terms of aberrant neurological impulses, evolutionary benefits, psychological projections, sociological insecurities, etc, etc. Theology, on the other hand, acknowledges the fragmentation of human nature and calls us to see things from a different perspective; it challenges us to ask new questions.

One change of thought that rises from this fragmented portrait of self is that human beings are not the coherent, 'rational animals' Aristotle believed us to be. In my readings in cognitive science I am struck by the number of times I have come across arguments relying on the premise that, as rational beings, humans cannot consciously hold both P and not-P. I beg to differ. It seems to me, as a student of theology and religious studies, that the ability to hold P and not-P is a universal one. Here, the sceptics may want to claim my observation as "proof" of the irrationality of religion, but I see irrationality, not as a property of religion, but as a property of human nature that appears to be essential for the development of culture. Rationality has proved itself a very useful tool, but a tool is all it is. I contend that as persons we are not fully and totally rational, instead we view fragmentary selves only through our fragmentary consciousness, and we become mesmerised by the illusion of a coherent, rational self somehow lurking inside.

So, returning to our WIIFM, how can these theological insights inform those in the field of artificial intelligence?

One: I believe we will know if and when we have achieved artificial consciousness when the intelligent machine we've created cannot be exactly reproduced. That is to say, we might reproduce the machine physically, bolt for bolt, circuit for circuit, but the reproduction will display a personality perhaps similar to that of its progenitor, but nevertheless distinct from it.

Two: a machine can be granted the status of person, if and only if its personhood is not fixed by a program, but is evoked through its relationship with its creators and other persons, whether they be mechanical and/or biological. In other words, the unique personality of the machine must grow and develop through its interactions with others.

And three: a conscious mechanical mind will demonstrate a fragmentary nature that is capable of irrational beliefs. Simulated consciousness will not be enough, for simulated consciousness is both programmed and rational. Only when a machine is capable of grasping beliefs that subvert its programming, and responds to stimuli in ways that are not always predictable, logical, or predetermined in some way by its programming, might we believe that we have created a machine in our image.

This third insight adds a cautionary note. I hesitate to include it because too often theology is perceived, perhaps somewhat unjustly, of forever pronouncing to the world, "Thou shalt not". So I'm hedging the caution inside what could almost be considered a theological inevitability.

To create a conscious machine in our own image would be to complete the theological circle, for our creation myths begin the theological enterprise by telling the story of how human beings were created in the image of God. It is hardly surprising then, that humans should desire to replicate God's creative actions through creating artificial intelligence. If we follow the theological circle round, the same Hebrew myths continue on to associate the human acquisition of consciousness with alienation from our creator. Christian theology tells us how, by creating us, God risked rejection, denial, and ultimately, crucifixion. Should we succeed in developing conscious, mechanical minds, our machines are set to become autonomous agents; the question we must ask ourselves us is this: Are we prepared to risk the same?

It is clear that theology has much to learn through listening to the sciences. As is evident at this conference, the field of Artificial Intelligence has already begun to enrich our notions of personhood. But dialogue must be two-way if it is going to continue and build relationships. If theology is to say anything to the world, we need to move the talks out of our own house. We need to have something to say, not just to our fellow theologians, and not just to our congregations. As theologians, we know the importance and value of theology, but for those outside the religious sphere, we must offer them a WIIFM. We must show them why it is worth their time and resources talking to us.