“Artificial intelligence, the technology that allows a computer to think like a human.” These are the opening words of a recent story by the BBC focusing on the impact of robots on religion. A fascinating topic which has been insufficiently discussed in the media.
Leaving aside the fact that artificial intelligence as it exists today, machine learning, neural networks and big data, doesn’t really simulate human thinking (a myth successfully debunked by Gary Marcus, Ernest Davis and Erik Larson), it is worth asking, nevertheless, what is or should be the link, if any, between artificial general intelligence (the only one that matters) and religious thinking, faith and practice. And, to my disappointment, the BBC piece doesn’t even scratch the surface. Unless by scratching the surface you mean asking what would people think about Siri/Alexa-type totems in places of worship. And the answers, at least the ones explored by the reportage in question, are trivial: people either like the idea of robot clergy (good for them!) or they object to it on two main grounds, one that the theo-robots are not credible because they have no soul (unlike a priest or a rabbi) or that it’s just plain weird. Both objections are objectionable: the soul is a philosophically problematic foundation to religious thinking and worship doesn’t always presuppose the existence of the soul; and the second, if people spent more time in the presence of robots, they might very well get used to them and embrace them.
Nor do I wish to be unfair. The 6-minute BBC video is hardly an opportunity to tackle this fascinating yet complex topic. But I believe more useful things could be said in less than 6 minutes. Let’s see if I can do that here.
One is indeed to say that we are far from modelling human thinking for the benefit of endowing machines with it. Artificial general intelligence, or real machine understanding, doesn’t exist and there is no reason to think that it should exist given the state of AI research and its direction of travel. Previous attempts at AGI have not so much failed than they’ve been abandoned in favour of narrow, unscalable applications of machine learning approaches. Alexa is impressive, but she’s totally stupid. From a human understanding point of view.
Another way is to ask what religious thinking requires and what it’s really about and see whether an artificial being, if it ever gets to the point of mastering human thought and understanding, could grasp religion in the true and deep sense of the word: an intuition of transcendence, a leap of faith into the unknown realm of belief. As far as this suggestion goes, I think literature, rather than science, provides useful insights.
Kazuo Ishiguro’s recent novel Klara And The Sun is about many things which have to do with AI, humanity and relationships, but I think it is, much more fundamentally, about the collision of faith and AGI. Klara is an artificial friend, an AF humanoid endowed with artificial general intelligence. She’s not a better version of Deepthink, Alexa or any other familiar iterations of machine learning concoctions. She’s real AI, and that’s why she’s compelling. She’s not Eugene Goostman, the fraudulent graduate of the Turing test. Klara’s understanding of the world she experiences develops an elusive poetic quality – though Ishiguro’s prose remains disarmingly terse.
Klara the AF ends up worshipping the sun and sacrificing herself for the person she serves – and saves. She doesn’t develop a religious language, the concept of god is not in her toolkit. Her neural network never brings the words theos and logos into a significant articulation. But she believes. Though she doesn’t really understand why or how. And she is acting religiously, and I’d argue, more humanly than most humans in the book, starting with the nihilist, agnostic doctor Capaldi and ending with Josie’s mother. Klara submits to Dr Capaldi plan to use her as Josie’s clone in the event of Josie’s impending death, but she never loses faith in the covenant she made with the sun to heal Josie. There is nothing artificial about the friendship proffered by this artificial friend. Klara’s ex voto is no-one but herself, a Christic figure crucified and discarded atop the yard for scrapped AFs.
Ishiguro isn’t bothered by the technical complexities of how Klara or her makers manage the jump from the visible to the invisible, from the light of the Sun hitting her sensors and powering her CPU to the cracks in the barn planks she so magnificently describes which allow the fading sunbeams to transform the barn into a place of worship, with Klara sitting in the inner sanctum, praying and asking deliverance for ailing Josie. The genius of the Klara hypostasis, if I may call it that, is that the robot is the only worshipper in a world where humans have become less human and faith an utter irrelevance. And this, to my mind, is the greatest question of all. For if humans locked themselves out of transcendence, would truly intelligent robots seek it out? Or, to use a biblical image, would an artificial being understand the power of the mustard seed and rush to find and plant one in the ground once the latter had become extinct?