Soul, Person, Dignity

Whoever says that a machine has no soul is asserting something very definite and at the same time something very indefinite. What is definite is the denial. What is indefinite is what is being denied. For the word soul does not mean the same thing everywhere in the sources of this book, and in most of them it is not explicitly defined at all.

Even the word itself is puzzling. No one knows for certain where the German word “Seele” comes from. A widespread interpretation connects it with “See”, lake, and understands the soul as “the one who belongs to the lake”, because according to ancient belief souls were thought to dwell in water before birth and after death. [Translator's note: German “Seele” means soul and “See” means lake; the interpretation mentioned here concerns the German word only.] The Latin “anima” means breath, a waft of air, and is related to the Greek word for wind. The Greek “psyche” is usually likewise associated with a waft of air and with breath; more recent research doubts this and suspects that the original meaning was the coldness of the corpse, which was transferred to the departing soul. What remains striking is how often the soul is connected with breath in the ancient languages, including in the Hebrew “nefesch”, to which the conclusion of this chapter returns. For human beings the soul was first of all what distinguishes a living body from a dead one. A machine does not breathe, and yet today people ask whether it could have a soul.

This chapter tries to clarify the word before the book continues to work with it. It asks about three closely connected concepts: the dignity that belongs to a human being, the person who remains the same over time, and the soul that, in many traditions, grounds both. Without this clarification, the thesis that artificial intelligence has no soul would be as empty as its denial.

Timothy Keller, pastor and author, begins with a definition on which, in his view, there is broad agreement. Human rights, he says, mean that every human being is entitled to be treated as a being of equal dignity and equal worth, simply because they are human, regardless of origin, sex, income, usefulness or talent.

His reasoning is theological. Every human being, he argues, is created in the image of God (imago Dei), reflects God and is therefore equally precious; the Bible therefore forbids not only killing a human being but even cursing one (James 3:9). This dignity, according to Keller, does not depend on capabilities. Whoever can no longer think or love, because of a disability or dementia, remains human and remains precious. Whoever ties rights to the capability for rational, free choice, he says, excludes infants and people with severe dementia.

Keller links this to a historical thesis. The idea of human rights, he argues, arose in the Christian Europe of the late Middle Ages from meditation on biblical teachings; he appeals to the political scientist Larry Siedentop and the historian Brian Tierney. At the same time he says that the idea arose only in the last two hundred years or so. The two stand in tension with each other, and there are competing derivations, from Stoicism, from the Enlightenment or from the experience of the Second World War. Keller's claim that no other culture knew a comparable doctrine is also disputed. The book therefore does not adopt the historical thesis but holds on to the systematic one: dignity that does not depend on capabilities needs a ground that does not lie in capabilities.

Enoch Sundaram, physician and preacher, describes the human being differently. As Chapter 3 has shown, he sees the human being as created threefold, physical, mental and spiritual, and he distinguishes the mind from the brain. The power that God has given the human being, he says, is above all the power to think. Sundaram identifies the place of the soul as an inner sanctuary, which he equates with the heart and the subconscious.

What is meant by the image of God needs explaining, for the word easily calls up false pictures. It does not mean that the human being looks like God. The Hebrew word “zelem” denotes a statue. In the ancient Near East kings had their statues set up in distant provinces; the statue represented the ruler where he himself was not present. In Egypt and Mesopotamia the king himself was also called the image of a god. The Bible transfers this title of honour to every human being, man and woman (Genesis 1:27). Not the king alone, but every human being represents God in the world and bears responsibility for it. This is why the image grounds dignity: whoever kills a human being lays hands on God's image (Genesis 9:6). The New Testament calls Christ the image of the invisible God (Colossians 1:15); in him it becomes visible what the image really means. In this understanding, dignity depends not on what a human being can do, but on what they stand for.

On one point Sundaram explicitly contradicts Keller. God, he says, created the human being in his own likeness; but after his disobedience Adam fathered a son who was like Adam himself and no longer like God. For Keller the image remains even after the Fall and grounds the dignity of every human being. For Sundaram the likeness is lost and must be restored through a renewal of thinking.

The two Christian voices of this book therefore do not hold the same doctrine of the human being. Keller grounds dignity in an image that remains. Sundaram describes the human being in terms of capabilities, above all thinking, and in terms of a likeness to God that has been lost. For the question of the machine this has consequences. Whoever defines the human being by the power to think must explain why a machine that thinks does not belong.

Chapter 8 presents the secular opposing view in more detail. Peter Singer, philosopher, ties moral status to properties such as sentience and self-consciousness, and calls preference based on species alone speciesism. For this chapter one point of his position is important, which Keller rightly emphasises: whoever ties dignity to properties cannot prevent human beings who lack these properties from falling outside. Singer himself draws this conclusion and does not consider an infant a person in every sense.

The opposing view is consistent, and it has a price. Keller's position is consistent too, and it also has a price: it cannot prove its ground, only confess it. Between the two lies the question that accompanies the whole book: is there a dignity that cannot be measured, and if so, by what does one recognise it?

The second question of the chapter concerns the person. What makes a human being the same over the years, although their body, their views and their memories change?

John Locke, a philosopher of the seventeenth century, gave an answer to this that still has an effect today. Personal identity, he held, depends on consciousness and on memory: I am the same as the one whose experience I can remember. Joseph Butler, theologian and philosopher, objected that this answer was circular, because memory already presupposes the identity of the person. Thomas Reid, philosopher, showed, using the example of an officer who as a general remembers his deeds as a young officer but no longer his childhood, that memory is neither necessary nor sufficient; personal identity, he held, is simple and cannot be broken down. More recent versions of Locke's theory speak of psychological continuity instead of memory and shape the debate to this day.

Sundaram stands, without saying so, in Locke's line. At death, he says, the entire memory, the whole of one's life experience, goes to God, who preserves it and gives it back with a new body at the Second Coming. What remains is memory. Keller is closer to Reid. For him the person themselves remains, transformed and yet recognisable.

Here the old question touches the new technology. If the person depends on memory, then the question arises why a system with memory should not be a person of this kind, and whether a stored memory of a human being preserves something of them. Sundaram answers this not via memory itself but via ownership: brain, mind and the capability to choose, he says, belong to God; the picture of life that the human being draws on them belongs to the human being. The memory of a human being is preserved by God, not by a machine. Reid's objection helps in a different way: if identity does not lie in storage, then not even the most complete storage preserves the person.

The Preface of this book names, besides the soul, two things that a machine does not carry: a biography and a mortality. Both can be grasped more precisely than it at first seems, and both are connected with the question of the person.

A biography is more than a record of events. It is a story that someone lives through, interprets and takes responsibility for, and that changes them. Aristotle tied practical prudence (phronesis), the faculty of doing the right thing in the particular case, to life experience, which young people still lack. Chapter 7 takes up this thought. A system can contain data about countless life histories. Whether it itself has a story that it has lived through is another question.

Mortality is more than an end. Hans-Georg Gadamer, philosopher, saw in the experience of one's own finitude a source of openness to others and to what one does not know. Keller describes death as separation from those one loves, and therefore as that against which the deepest hope is directed. Sundaram describes how the body returns to the earth. In all three descriptions death is something that gives life weight, because it cannot be repeated.

A machine can be switched off, but in this sense it cannot die. It can be copied, backed up and restarted. With a machine, what once was is not irretrievable. It does not yet follow from this that it therefore lacks a soul. But it does follow that it lacks an experience that, in the traditions of this book, is closely connected with the soul: the experience of having only one life.

Geoffrey Hinton, computer scientist, one of the founders of machine learning with neural networks and awarded the Nobel Prize in Physics in 2024, sees precisely in this not a deficiency but a superiority. When a human being dies, their knowledge dies with them. A digital intelligence, by contrast, can be transferred to new hardware as long as its connection strengths are stored, and many copies can share with one another what each one has learned. Such systems, he says, are immortal; they have solved the problem of immortality, though only for digital things. Hinton and the traditions of this book describe the same finding and assess it in opposite ways. For Hinton the absence of finitude makes the machine stronger. For Gadamer, for Keller and for the psalm that asks to learn to count one's own days (Chapter 3), it is precisely finitude that makes the human being prudent. Which assessment holds cannot be measured; it belongs to the assumptions of which Chapter 1 speaks.

With these distinctions it is possible to set in order what can be meant when the soul is spoken of. In the material of this book and in the debate about artificial intelligence at least five meanings can be distinguished.

First, the soul as relationship. For Keller the soul is not a substance but the person in their relationship to God, the person who loves God and is loved by him. In this sense a machine has no soul, because it does not stand in this relationship. The statement is theological and cannot be tested empirically.

Second, the soul as the innermost. For Sundaram it is the inner sanctuary, the place where God wants to dwell and where conscience speaks. Here too a machine has no soul, because it has no such place. Whoever, however, equates the innermost with the subconscious, as Sundaram does, makes it vulnerable to the question of whether machines too might not have hidden processes that determine their behaviour.

Third, the soul as memory and life story. Then the question of whether a system with memory has a soul is not a meaningless question, and the answer depends on the objections of Butler and Reid from section 6.5. Butler countered that memory already presupposes the identity of the person instead of grounding it. Reid showed that memory alone leads to contradictions: an old general remembers his time as a young officer, the officer remembers his childhood, but the general no longer remembers the child. According to Locke he would then be at the same time the same person as the child and not the same person. If both objections hold, memory alone does not yet make a person, not even in a machine.

Fourth, the soul as the totality of mental functions. This is the functionalist reading that Chapter 3 introduced and Chapter 8 develops further. If soul means only perceiving, thinking, feeling and willing, then it is not bound to biology, and in machines it is not ruled out in principle.

Fifth, the soul as an empty word. Some consider the concept superfluous, because it explains nothing that could not be explained without it. Their best-known proponent today is the Israeli historian Yuval Noah Harari. At the World Economic Forum in Davos in 2020 he said that human beings were now hackable animals. Harari also held at the time that the idea that the human being has a soul or a spirit and a free will, and that no one knows what goes on inside them, was over. Whoever has enough data about a human being, according to Harari, knows them better than they know themselves. Then the question of whether machines have a soul is wrongly posed, for human beings as for machines.

Alongside these stands a possibility that is seldom voiced. Whoever believes in a Creator cannot simply settle by definition the question of whether God could give a soul to a creature, even to one that human beings have built. That is a theological question which the book must leave open.

In the debate about artificial intelligence, the question of dignity has taken on a harsh form. In a panel discussion recounted in section 9.6, one participant, presumably Roman Yampolskiy, weighed individual deaths against eight billion human beings and all future generations and declared them not significant. Another participant insisted that this is about people with names.

This chapter can say what is at stake in this statement. Whoever ties dignity to the individual, independently of their usefulness for the whole, cannot offset that individual against a sum. Keller does this with the image of God. Sundaram measures the strength of a Christian congregation by how gently it deals with the worst sinner. Whoever, by contrast, ties worth to consequences for the whole can offset individuals, and sometimes even must. The question of the soul is not an academic one here. It decides whether a human being is a quantity in a calculation or something that eludes calculation.