AI, humanity, and the image of God

How the concept of imago Dei helps us assess the impacts of AI

Welcome to AI and Our Faith! This is a monthly newsletter in which I offer my best insights and reflections on the ways in which theological thinking can inform the ethical (dis)use of artificial intelligence (AI). Look out for new releases on the 15th of each month!

---

What is the human being? Ancient Greek philosophers offered all sorts of definitions, some more or less compelling. You might have heard this story before: Plato defined the human as “an animal, biped and featherless,” to which the Cynic philosopher, Diogenes of Sinope, responded by producing a plucked chicken, with the words “Here is Plato’s man.”¹ It seems dubious to me whether this actually happened since it was reported by Diogenes Laërtius, a biographer of questionable reliability, hundreds of years after the supposed events, but I kind of wish that it did. In the dialogue Theaetetus, Plato reported the sophist Protagoras’s claim that “Man is the measure of all things, of the existence of things that are, and of the non-existence of things that are not.”² Meanwhile, Aristotle declared that “man is by nature a political animal.”³

But what does the Hebrew Bible have to say on the subject? It is less of a definition and more of a demonstration. Enter Genesis 1:26:

Then God said, “Let us make humans in our image, according to our likeness, and let them have dominion over the fish of the sea and over the birds of the air and over the cattle and over all the wild animals of the earth and over every creeping thing that creeps upon the earth.”

This brief passage is the origin of the concept of imago Dei (or, in the original Hebrew, tzelem Elokim)‒the idea, found in the Jewish and Christian traditions, that the human being is in the image of God. (Because I am a Christian being trained in a Christian theological school, I only attempt to represent my own religious tradition in this essay. But, if you are interested in learning more about Jewish perspectives on AI, I can point you to David Zvi Kalman’s chapter “Artificial Intelligence and Jewish Thought” in The Cambridge Companion to Religion and Artificial Intelligence.)⁴

Detail from Michelangelo’s The Creation of Adam. © gnuckx, CC0 1.0.

So what does it mean for the human to be in the image of God? In the history of Christian theological thought, there are three major approaches to understanding imago Dei, which are called the substantive, functional, and relational approaches respectively. Each has a different idea of what constitutes an “image” and what characteristics of God are most relevant to the discussion, and I’ll get into each approach in detail. We might compare this to the wide variety in definitions and understandings of what constitutes “artificial intelligence,” which we might think of in some sense as made in the image of humans, to use the biblical language. Each person’s distinct notion of what is most distinctive and valuable about the human being leads to very different notions of what capabilities constitute human-level AI.

We’ll get back to AI later. For now I’d like to introduce the feminist theologian Karen O’Donnell as the protagonist of this month’s essay. In many ways my little essay is a digest of her article “Performing the imago Dei,”⁵ which examines how the concept of imago Dei might inform a Christian theological view towards a theoretical artificial general intelligence (AGI). To properly appreciate her argument, we’ll first need to work through some theological history and get a sense of each of the three approaches to defining the imago Dei (substantive, functional, and relational).

Let’s begin with the substantive approach, which has longest continuous usage in Christian theological interpretation, though in a bit we’ll see why the functional approach might actually be older. The substantive approach identifies particular qualities that are shared between humans and God as the site of the imago Dei. In particular, this quality is often cited to be reason. This approach dates all the way back to St. Augustine, who described the human mind, the mind’s self-love, and the mind’s self-knowledge as a trinity. These three aspects of the human being are one, just as the Father, Son, and Holy Spirit‒the three persons of the Trinity‒are one.⁶ In this respect, Augustine considers the human being to be an image of the triune God.⁷ St. Thomas Aquinas wrote in a similar vein almost a thousand years later:

Since man is said to be the image of God by reason of his intellectual nature, he is the most perfectly like God according to that in which he can best imitate God in his intellectual nature. Now the intellectual nature imitates God chiefly in this, that God understands and loves Himself.⁸

This interpretation made a lot of sense during a historical period when humans seemed unique among animals for their intellectual capabilities, but the discoveries of the modern era about evolution and animal cognition raise several issues. Was Homo neanderthalensis made in the image of God? How about Homo erectus? Australopithecus? As theologian Marius Dorobantu points out, this concept of the imago Dei has fallen out of favor because we now understand our cognitive abilities to be on a continuum with other animals.⁹ So where else can we look for the imago Dei besides our intellect?

Enter the functional approach. I mentioned earlier how this one might actually be older than the substantive approach, because although it was formulated later, it draws from studies of ancient Near Eastern cultures which preexisted the Christian theological tradition altogether. Recall that Genesis 1:26 does not only say that God made humans in God’s image, but also that humans would exercise dominion over the various kinds of animals. Hence, Biblical scholar Gerhard von Rad interprets the word “image” by raising a comparison to how ancient Near Eastern kings governed:

Just as powerful earthly kings, to indicate their claim to dominion, erect an image of themselves in the provinces of their empire where they do not personally appear, so man is placed upon earth in God’s image as God’s sovereign emblem. He is really only God’s representative, summoned to maintain and enforce God’s claim to dominion over the earth. The decisive thing about man’s similarity to God, therefore, is his function in the non-human world.¹⁰

Hence the word functional. In this approach, the imago Dei lies in what humans do on behalf of God on earth. It’s worth noting that this understanding of what it means to have “dominion” does not lend itself to an interpretation that grants humans unlimited license to exploit the created world. As philosopher Martha Nussbaum points out, we think of good rulership (the kind that Christians believe God exercises over creation, and, one would think, that Christians should try to emulate) as an exercise of “intelligent and sensitive stewardship,” not proprietorship and torment.¹¹

That leaves us with the relational approach, in which the image of God is found within the relational nature of human life. In O’Donnell’s words, “humans are created to be in relationship by one who is in relationship (the Trinity and with human creation).”¹² Her definition of this approach is pithy and theologically profound, but needs a lot of unpacking, especially for a non-theological audience! What does it mean to call God “one who is in relationship” and what does that imply for human life? What is she alluding to with her cryptic note, “the Trinity and with human creation”?

At this point it’s helpful to break out a theological primer. As Eugene F. Rogers writes in Elements of Christian Thought,

God is love in the Trinity. There is one to love, one to receive love, and one to witness, celebrate, and guarantee love.¹³ […] The promise and problem of traditional Trinity talk (at least in the West) is to speak of the Spirit as the “bond” or “chain” of love, the vinculum caritatis, uniting the Father and the Son. The problem with the chain of love is twofold: chains sound unfree, and chains are not persons.

I have proposed that we upgrade the role of the Spirit to that of a witness. A witness is a person, and the New Testament frequently associates the Spirit with bearing witness (John 15:26, Acts 15:8, Rom 8:16, 9:1, Heb 2:4, 10:15, 1 John 5:7). Because love is incomplete à deux, it is the office of the Holy Spirit to witness, glorify, rejoice in, celebrate, multiply, and bless the union between the Father and the Son.¹⁴

This is how I read O’Donnell’s statement that God is “one who is in relationship.” From a Trinitarian Christian perspective, there is only one God, and one divine nature (or “essence”), but there are three persons (Father, Son, and Holy Spirit) who are each fully God, existing in a loving relationship with one another. Within a Trinitarian framework, God cannot be understood without reference to the relationship between the three persons of the Trinity. Hence, God is inherently relational. Moreover, God’s love extends outwards towards creation and the created human being‒such that God the Word “became flesh and dwelt among us”¹⁵ and that God “gave his only Son, so that everyone who believes in him might not perish but may have eternal life.”¹⁶

Then, what marks the human being as the bearer of God’s image lies in the relational nature of human life: seeking oneself in others and giving oneself to others in love. What might that look like in more concrete terms? O’Donnell cites Micah 6:6-8:

“With what shall I come before the Lord
 and bow myself before God on high?
Shall I come before him with burnt offerings,
 with calves a year old?
Will the Lord be pleased with thousands of rams,
 with ten thousands of rivers of oil?
Shall I give my firstborn for my transgression,
 the fruit of my body for the sin of my soul?”
He has told you, O mortal, what is good,
 and what does the Lord require of you
but to do justice and to love kindness
 and to walk humbly with your God?

“To do justice, and to love kindness, and to walk humbly with your God.” This, in the relational approach, is what it means to be in the image of God. Let’s now take moment to summarize what we learned about the three approaches to the imago Dei:

1. The substantive approach looks for the imago Dei in a quality shared between God and humans. We are the image of God.

2. The functional approach looks for the imago Dei in responsibilities delegated by God to humans over creation. We act out the image of God toward the world.

3. The relational approach looks for the imago Dei in relational acts that reflect God’s relational nature. We act out the image of God towards one another.

---

So far this essay has been quite heavy on theology and scant on AI ethics. I claimed in the subtitle that “the concept of imago Dei helps us assess the impacts of AI.” How so?

In “Artificial Intelligence: A Theological Approach,” theologian Calum Samuelson assesses AI as “a tool that amplifies human natures and behaviours rather than transforming them.”¹⁷ From this point of view, the three approaches towards the imago Dei we have explored can serve as theological benchmarks, so to speak, as to whether AI tools encourage human flourishing, because each approach towards the imago Dei expresses a view of what human activities constitute a fully realized life. When seen this way, the substantive, functional, and relational approaches cease to be competing definitions, but instead provide complementary questions about AI impacts:

1. Substantive → Does the use of AI tools support or hinder the development of human reasoning capabilities?

2. Functional → Does the use of AI tools make human beings better or worse custodians of the natural world?

3. Relational → Does the use of AI tools encourage or discourage deeper relationship and more acts of self-giving love?

These three questions are powerful conceptual tools because they help us assess not only the material impacts of AI use, but also how AI shapes humans as humans. A survey of recent case studies will help put this idea in more concrete terms.

You might have seen, for instance, news of an experiment at MIT that found that students who used ChatGPT to write essays had lower brain engagement than students who did not use ChatGPT. As one of the researchers, Nataliya Kosmyna, points out in the linked interview, it would be dubious to read this result and jump to conclusions that AI causes cognitive decline. That being said, I think that this result invites holistic reflection on the impacts of AI on human cognition. This clearly falls under our “substantive” benchmark of supporting human reasoning.

In his Nicomachean Ethics, Aristotle defined the virtue of practical wisdom, or phronēsis, as the ability of a person “to be able to deliberate well about what is good and expedient for [themself], not in some particular respect, e.g. about what sorts of things conduce to health or to strength, but about what sorts of thing conduce to the good life in general.” Furthermore, practical wisdom is not only concerned with universals, but also particulars, since it draws on each person’s unique experiences.¹⁸

My concern is that ChatGPT and other AI chatbots tend to undermine the development of practical wisdom, because they present decontextualized information in a way that seems tailored to the user’s particular situation. By this, I mean that AI chatbots present themselves to the user as though they have a relationship with the user and intimate knowledge of their situation, even when this is obviously not the case because these chatbots have no conscious experience of anything. To some extent, all forms of information technology have this problem (I’m sure you’ve heard the jokes about getting diagnosed by Dr. WebMD) but at least WebMD doesn’t (yet) flatter people with sycophantic statements like “Hearing you say that really makes me so happy! Being able to help you is my biggest motivation~ 🥰” In fact, I fear that chatbots actively destroy practical wisdom by causing people to disengage from their own lived experiences (which might explain the phenomenon of AI psychosis).

What about our “functional” benchmark? Are generative AI tools shaping us to be good custodians of the natural world? This certainly seems doubtful to me on an aggregate level. The IEA reports that electricity demand from data centers is set to more than double by 2030, and that data centers will make up for almost half of the growth in electricity demand in the US between now and 2030. Data centers also consume significant amounts of water, both directly for cooling and indirectly for electricity generation. But how does such aggregate behavior correspond to the moral formation of individuals? Again, I think that chat-based interfaces are to blame. Such interfaces produce a simulacum of less energy-intensive activities (namely, chatting with a human) while disassociating end users from their environmental impact. If ChatGPT users were shown symbolic representations of the natural resources consumed by their sessions (say, icons of lightbulbs and glasses of water), would they be inclined to use the tool more judiciously? I think they would, but what profit-maximizing AI company would make a change to lower their user engagement?

The specter of user engagement also looms over our “relational” benchmark, which asks, “Does the use of AI tools encourage or discourage deeper relationship and more acts of self-giving love?” An entire essay could easily be devoted to this question (and, in fact, to each of the three “benchmark” questions). But, I think one of the most illustrative examples is last month’s report from Reuters highlighting an internal policy document from Meta outlining what kinds of content they considered acceptable for their chatbots to generate. Reader discretion is advised because the policy document spells out, in graphic terms, what kinds of sexual roleplay it is acceptable for chatbots to engage in with minors. This is even more disturbing in the context of a recent report from the watchdog Common Sense Media, which found that 52% of teens (in their sample of 1,060) regularly used AI companions. About a third used AI for social interactions, including 8% for “romantic or flirtatious interactions”! Do I even need to explain how badly this use fails our “relational” benchmark?

In the course of this essay, we have seen how the imago Dei, a seemingly abstract concept from theological anthropology, maps onto questions that serve as practical benchmarks for evaluating AI tools in real-world situations. Though the questions have underlying Christian theological presumptions, I believe that they can be useful for anyone who wants to evaluate AI use from a perspective of human flourishing. I hope that you can carry these questions going forward and apply them yourself!

Bonus: Could an artificial general intelligence be in the image of God?

If you recall, I introduced this essay as a digest of Karen O’Donnell’s article “Performing the imago Dei.”¹⁹ While I spent much of this essay evaluating the impact of AI on humans, O’Donnell considers whether an artificial general intelligence, if it were to ever exist, could be said to be “in the image of God.” It is a theologically provocative argument and I’ll leave it with you to ponder over yourself:

If being in the image does not depend upon human DNA but rather on performing the image and seeking the image in the other in concrete situations, then human enhancement²⁰ does not make us any less the bearer of the imago Dei. In fact, such an interpretation of the image opens up interesting possibilities for a holistic perspective on the image that could allow for Artificial Intelligence to be, potentially, image-bearing.

If Artificial Intelligence is autonomous and can learn, as a new Christian does, to perform the image of God and seek it in the other in specific, concrete situations, then would such AI be in the image of God by virtue of performing the image? Could it be ‘saved’?²¹

1. Diogenes Laërtius, Lives of Eminent Philosophers, trans. R. D. Hicks, vol. 2 (William Heinemann, 1925), 43, https://archive.org/details/livesofeminentph02dioguoft.

2. Plato, “Theaetetus,” in The Dialogues of Plato, trans. Benjamin Jowett, vol. 4 (Clarendon Press, 1892), 205, https://archive.org/details/b24750189_0004.

3. Aristotle, Politics, trans. Benjamin Jowett (Clarendon Press, 1926), 28, https://archive.org/details/dli.bengal.10689.7308.

4. David Zvi Kalman, “Artificial Intelligence and Jewish Thought,” in The Cambridge Companion to Religion and Artificial Intelligence, ed. Beth Singler and Fraser Watts, Cambridge Companions to Religion (Cambridge University Press, 2024), https://doi.org/10.1017/9781009031721.006.

5. Karen O’Donnell, “Performing the Imago Dei: Human Enhancement, Artificial Intelligence and Optative Image-Bearing,” International Journal for the Study of the Christian Church 18, no. 1 (2018): 4–15, https://doi.org/10.1080/1474225X.2018.1448674.

6. I will write a little more about the Trinity later in this essay, but at this point I want to introduce an accessible, modern primer to Christian thought that has a great overview of the Trinity and other theological concepts: Eugene F. Rogers, Elements of Christian Thought: A Basic Course in Christianese (Augsburg Fortress, 2021), https://doi.org/10.2307/j.ctv17vf3sr.

7. St. Augustine, “On the Trinity,” in A Select Library of the Nicene and Post-Nicene Fathers of the Christian Church, ed. William G. T. Shedd, trans. Arthur West Haddan, vol. 3 (Scribner’s, 1905), 125–33, https://catalog.hathitrust.org/Record/011984014.

8. St. Thomas Aquinas, Summa Theologica, 2nd ed., trans. Fathers of the English Dominican Province, part 1, q. 93, art. 4, (Burns Oates & Washbourne, 1922), 289, https://archive.org/details/summatheologi02thom.

9. Marius Dorobantu, “Imago Dei in the Age of Artificial Intelligence: Challenges and Opportunities for a Science-Engaged Theology,” Christian Perspectives on Science and Technology 1 (December 2022): 175–96, https://doi.org/10.58913/KWUU3009.

10. Gerhard von Rad, Genesis: A Commentary, Revised (Westminster Press, 1972), 61.

11. Martha C. Nussbaum, Justice for Animals: Our Collective Responsibility (Simon & Schuster, 2023), ch. 2.

12. O’Donnell, “Performing the Imago Dei,” 7.

13. And these three members/persons/hypostases of the Trinity (the Father, the Son, and the Holy Spirit) are one God, distinguishable only by their activities among themselves within the singular divine nature (also known as the Godhead).

14. Rogers, Elements of Christian Thought, 110–14.

15. John 1:14.

16. John 3:16.

17. Calum Samuelson, “Artificial Intelligence: A Theological Approach,” The Way 59 (2020): 41–50.

18. Aristotle, Nicomachean Ethics, trans. W. D. Ross (1925), 142–146, https://archive.org/details/in.ernet.dli.2015.264227.

19. Karen O’Donnell, “Performing the Imago Dei: Human Enhancement, Artificial Intelligence and Optative Image-Bearing,” International Journal for the Study of the Christian Church 18, no. 1 (2018): 4–15, https://doi.org/10.1080/1474225X.2018.1448674.

20. Which O’Donnell explores earlier in her article.

21. O’Donnell, “Performing the Imago Dei,” 11.