CAN these bones live? Ever since Pygmalion, we’ve been willing our statues to life: Pinocchio, Frankenstein’s Monster, the Snowman. These days, the breath that we give them is Artificial Intelligence (AI), and it’s developing fast.
Just in the past few months, an AI-generated picture has won an art competition at the Colorado State Fair; and the release of ChatGPT has enabled weary academics to outsource the creation of their curricula and model essays to AI. After AI successes in chess and the Chinese game of Go, in November the program Cicero also achieved human-level performance in the American board game Diplomacy; and, in December, DeepNash mastered the board game Stratego.
That’s the Christmas board games and the thank-you letters sorted. Perhaps next year, Alexa will get your Aga to make Christmas dinner for you, while the festive drones deliver your presents, and you relax in front of a personalised movie.
Ours is the generation in which science fiction may finally become science reality, as these tools become more and more integrated into our daily lives. It happened with fire and with the wheel; it happened with medicine and electricity; and now it is happening with AI and robotics. And, while humans have always used tools as enhancements and to save time, this may be the first time that we have set about building something that is designed to replace us.
It is interesting that we call it artificial “intelligence”. Perhaps the experience within living memory of the horrors of eugenics drove us to focus far more on the creation of artificial minds than artificial bodies, particularly as moral and legal scruples still protect us from human cloning.
But, at this time of year, it is the body that is the star of the show, because God chose to become incarnate as a physical human baby. So, how are we getting on with artificial bodies, albeit of the non-biological kind?
IN 2017, in a lab in Columbia, Hod Lipson was working on a simple robot: four small mechanical legs, given the task of moving from one side of the room to the other. It was programmed using Deep Learning: using artificial neural networks and reinforcement learning so that the robot could teach itself.
After a few days, the robot had learned how to take its first steps, and had crossed the room. In getting the robot ready for a formal demonstration, the team examined how it was using its neural network to learn, and they discovered something unexpected: the robot had also taught itself how to read their facial expressions. Realising that observer feedback is data, it had dedicated a neuron to processing this information. Humans instinctively recognise this choice: we have all stood around cheering on a toddler, because our feedback helps them learn how to walk, too. Clever!
Fear not; for we are still a very long way away from replacing your beloved King’s choristers with singing androids on Christmas Eve. But it seems to me that we are far enough down the line of creating artificial minds and bodies that we ought at least to check in with God about it a little more than we seem to be doing. If God made us in his own image, and himself became human, what ought we to learn from what we can discern about his blueprint for us?
THEN we start to notice something a bit fishy about AI. It is not copying human design: it is copying only those bits of it that the programmers reckon pass muster. The rest — all that junk code — is deliberately kept well out of the way, so that our creations are not sullied by it.
But, if you zoom in on this spurned code, you find the flotsam and jetsam of soul: free will, emotions, intuition, uncertainty, mistakes, meaning-making, and storytelling. When you consider why God designed in these seemingly irrational properties, you glimpse something that resembles a suite of ameliorators with a common theme.
Free will gives us a feeling of agency and purpose. Our emotions promote healthy relationships and community living, and we use our intuition and conscience to avoid danger and mistakes. Uncertainty keeps us seeking out others for guidance, and we hold our communities together through meaning-making and storytelling. Together, this “junk” code promotes co-operation and the kind of reciprocal altruism which creates communities that thrive.
But, when it comes to designing AI and robots, we are not extending the courtesy of our own design to them. In fact, they are deliberately designed to be individualistic and quite brilliant psychopaths. We had thought that that would serve our purpose in creating them, but now I am not so sure.
Perhaps it is Christmas whimsy to suggest a gift of soul, even if that were in our gift. But could we at least look at their design again, to see if making AI more human might actually be better than making it less?
Dr Eve Poole writes on theology, economics, and leadership. Her book Robot Souls is due to be published by Routledge in 2023.