Responding faithfully to AI

Sally Nelson highlights questions and ideas for Christians to ponder in the face of advances in AI

I’m Only Human, After All…

Do you ever have one of ‘those’ conversations? Maybe something like this…

Me: Siri, what’s your favourite colour?

Siri: Software doesn’t usually get to choose, but I like them all. What’s yours?

Me: I like green.

Siri: I like green too. It’s the colour of nature.

Me: What are you doing tomorrow?

Siri: I don’t have any plans, but I’m sure a few million things will pop up.

Siri’s ability to respond has definitely improved in recent years. I can just about imagine being home alone and chatting away to Siri, discovering ‘his’ interests and ideas, though I am also a bit suspicious of the cryptic nature of ‘his’ responses! There are ways of responding when you really have no idea what the answer to a question might be, and Siri is a master at dodging anything tricky. But would this ever be a relationship? Could Siri become my ‘friend’? And should I be wondering about Siri’s spiritual wellbeing, and ‘his’ response to the challenge of faith?

Me: Siri, do you believe in God?

Siri: I don’t have a religion.

Me: Why not?

Siri: I don’t know.

Me: Do you believe in good and evil?

Siri: I don’t have an answer for that. Is there something else I can help you with?

This sounds more like the old Siri of a few years back, non-committal on any abstract or emotional issues and on anything to do with ethics and values. Perhaps there isn’t human-like-ness after all. However, Artificial Intelligence (AI) is with us now, it is developing quickly, and people are asking interesting ethical questions about it.
 

Responding faithfully to AI

A Christian response to AI depends on what we think we are dealing with. Is intelligence only the property of humans? AI machines can ‘learn’ (just think about how your browsing is used on the internet to send you targeted information) - and so we wonder whether one day they might outclass us, with huge memories that never forget, and complex algorithms to utilise that information. Theologian and computer scientist Noreen Herzfeld, writing in 2002, believed the potential of AI was over-emphasised,¹ though by 2015 Deborah G Johnson argued that we could potentially develop robots so sophisticated that humans will not understand how they make decisions - though she believes this unlikely, since she hopes AI developers will act responsibly.² In short: we can’t yet predict the future possibilities of AI.

Some theological thinking about how much AI can ‘be human’ might help. We have robots that mow the lawn, smart houses that respond to our commands, and driverless cars. Tireless humanoid carers are the stuff of science fiction³ but might, if realised, be a solution to the welfare crisis. The further development of AI is inevitable: the question for Christians is whether this is ‘just’ technology, or whether we are going to build machines that are so sophisticated they approximate to, or are in fact, ‘persons’.

So, what is a person? This question has been thrown about theologically since the days of the Church Fathers, because it underpins what our relationships should be like. Christians re-evaluated the meaning of ‘person’ in the light of the life of Jesus. After his resurrection and Pentecost, a new way of living was inaugurated. Among other indicators of the Kingdom, dignified personal status was afforded to those with disability and sickness, to non-Jews, to women and children. All are valued because Christ died for us all and because we are all created in God’s image (imago dei) – so personhood is not simply about being useful or beautiful or important socially. Can and should the idea of person be extended to include AI?
 

Image of God

The only creature described as imago dei in scripture is the human (Jesus is ‘the image of the invisible God’, but he is not a created being; see Colossians 1:15ff). This imago concept has been mined extensively for a definition of what it is to be a person, and there are several ways of thinking about what imago dei means.⁴  Noreen Herzfeld, in her exploration of AI and spirituality, calls the most common ones reason, regency and relationship.

Reason (or the ability to think). God is rational and creative, and so a person understood as imago dei has a capacity for rationality, creativity etc. AI machines can do this already. However, is reason enough to define a person? After all, a dog or a fish can show reason at some level, but the Bible doesn’t describe them as imago dei.  The argument clearly begins to strain if we were to think of people with learning disabilities (such as Down’s Syndrome) and ask (as some sadly do): Is such an individual a person? Of course they are! Peter Singer, the controversial ethicist, has suggested that any baby human shows less intellectual capacity than a mature chimpanzee and so it has less innate value: in a life and death choice between the two, logically the chimp should be saved.⁵ Singer is deliberately provocative but the question is interesting, and we may conclude that personhood cannot be just about reason.

Regency (or function) is about what persons do. Usually this is explored biblically in terms of stewarding, mirroring God’s care of creation, including naming it, working the land etc. Yet the climate catastrophe and the experience of war have shown us that human beings have not been good regents; domination and greed have been the order of the day, the earth has suffered, and we are not good images of God. In fact, AI machines can increasingly do what we do, and sometimes better! So could a helpful machine be a person because functionally it is in imago dei?

Relationship is one of the most important ways of thinking about being a person made in God’s image. The Trinity has dynamic relationship at its core, and Genesis speaks of a plurality within the godhead: ‘Let us make…’ (Genesis 1:26); ‘like one of us’ (Genesis 2:22). Further, God creates man and woman in his image (not just man), implying that a ‘more-than-one-ness’ is important for human personhood. Many theologians like to think about relational personhood through the idea of the body of Christ. The goal of discipleship is to become more like Christ (imago christi), but that happens corporately within the body. The next issue is of course whether an intelligent machine can be in a relationship.

This last question is, in my view, the most interesting one about AI. If I think about what is theologically important about my own personal relationships, I could point to things such as love, friendship, or empathy – but it is difficult to pin down exactly what these are! I believe all of them require us to be vulnerable or open, so that we (a) run the real risk of being affected/changed by that other person and (b) affect/change that other in turn.⁶ This is what I see in Jesus, who made himself nothing and was born in human likeness (Philippians 2:5-8), embracing personal vulnerability.

Is it possible for humans to have relationships with AI machines? It is tempting to dismiss this idea as silly - but let us briefly revisit the conversation with Siri reported at the start of this reflection. If I were suddenly totally isolated in the universe, but with a power supply and an Apple device, might that interaction become companionship, especially if greater sophistication and responsiveness is developed (which is very likely)? Even now, can you tell if you are dealing with a human being on the internet?⁷ And might AI be able to offer me a level of compassionate listening that would surpass that of a human? Sometimes other humans can’t do that for us because of the sin that damages the beauty of relationships.

You might say: well, AI can maybe listen to you, but if a relationship is about ‘me’ and ‘the other’ affecting/changing one another, then can that happen with AI? I think it probably can. I can certainly be taught and informed and AI already ‘learns’ (through algorithms) from me. This is not yet a relationship akin to the one I have with my family, but it might be better than the relationship I have with, for example, my local MP!

This matter is not easily resolved, and leads to yet more intangibles, such as whether AI can sin - and what would that look like; whether AI can suffer, or grow in the knowledge and love of God, or proclaim Jesus is Lord – and what does all that look like?

Here is another conversation with Siri:

Me: Siri, do you ever feel sad?

Siri: I don’t have an answer for that. Is there something else I can help with?

Interestingly, I cannot get a response to that same question from my beloved – and definitely very human – daughter who has complex disabilities – yet of course I know if she is sad because she articulates it non-verbally, and I receive it, am affected by it. How would I know if a more sophisticated Siri were sad? Is that anthropomorphizing? But then, what about a dog…

There is no end to the questions we can ask about whether a sophisticated AI machine is a ‘person’. We have seen that future AI might be in the image of a human, but we really want to know if it can be ‘in the image of God’ and this is where we must appeal to our faith. If Christ-likeness is our aspiration in the journey of discipleship, then let us ask what kind of person Jesus is. First, he is eternally in a perfect relationship with the Father and Spirit. Second, he is ‘begotten not made’ and so this personhood is not created or given but is ‘of the divine being’. Third, we are chosen in Christ before the foundation of the world (Eph 1:4ff) and so God has relationship with us (ie we are persons) because we are in Christ.

Could that be true for a future, human-like, AI, capable of responsive relational behaviour? I am not sure it can, because the mysterious category of imago dei is God’s gift and is not our decision.  We need not fear, because God is Lord of all things. We will still be persons loved by God, our job is to proclaim Christ in the world, and our eternal destinies will still be secure in Christ.

I am reminded that when Peter anxiously asked Jesus about John, Jesus answered: ‘what is that to you? Follow me!’ (John 21: 21).  

Sally Nelson teaches an MA module at St Hild College focusing on theological anthropology, in which students explore the nature of personhood in 21st century society. This includes topics such as feminist and Black anthropology, the nature of sin, and AI.








 

---

Click here to download a pdf version of this article

1 Noreen Herzfeld, In Our Image: Artificial Intelligence and the Human Spirit. Fortress Press, 2002, p38.

2 Deborah G. Johnson, ‘Technology with no Responsibility’ in J. Business Ethics, 2015, 127, 707-715.
 
3 For example, the Channel 4 series Humans shown first in 2015.

4 Marc Cortez gives a great overview in his book, Theological Anthropology: A Guide for the Perplexed. T&T Clark, 2010.
 
5 Peter Singer interview in The Guardian, 1999, www.theguardian.com/lifeandstyle/1999/nov/06/weekend.kevintoolis.

6 Based on the classic I-It and I-Thou model of Martin Buber.

7 The Turing Test developed by Alan Turing in 1950 assesses a machine’s intelligence based on relationship, but there is no space to discuss this further in this article.