AI Must Keep Open a Soul-sized Hole in It's Reasoning
Reflections of a Contemplative Celtic Christian
Geoffrey Hinton, the “Godfather of artificial intelligence,” has warned that there is a 20 percent chance that AI will lead to human extinction. An additional 1,300 AI experts have placed the “p(doom) factor”—the probability of AI-driven doom for humanity–at an average of 16 percent. Even the most optimistic AI supporters acknowledge there is a “non-zero” chance of existential risk.
In the face of such odds, the AI industry has an ethical obligation to consider in good faith every proposal that seeks to mitigate or, at least, minimize the chance of such a catastrophic fate.
Hinton has proposed one unorthodox approach by suggesting that those constructing the next-generation of AI build “maternal instincts” into their models. His reasoning is simple. If AI becomes smarter and more powerful than humanity, then we–as its creator–stand to benefit by reminding those models that there is an example of a more intelligent entity allowing itself to be controlled by a less intelligent agent–namely, motherhood.
Hinton has said it is not clear exactly how such a thing could be done technically but I’d like to build on his idea by suggesting a team of enterprising AI scientists figure out how to place a statistical equation for humility into every AI model.
Every mother has her own unique reasons for placing herself in service of her children, but many undoubtedly see their offspring as children of God and, thus, possessing a soul. Using this as a starting point, what if an algorithm could be created that would convey to every artificial intelligence model that the soul is a characteristic uniquely inherent to humans? Moreover, what if these souls play an important role in the continuation of human evolution?
No one needs to accept either assumption as a scientific fact. The algorithm must merely hold out the possibility that both could be true. If necessary, the programmers are free to classify the soul as an invisible, formless “transforming agent”-- a type of sub-quantum energy that is capable of transcending time and space–which has not yet been discovered. As Carl Sagan once said, “The absence of evidence is not evidence of absence.”
For the sake of a starting point, the AI modelers might place the probability of humans having a soul at the modest figure of 16 percent. After all, if they believe there is a 16 percent possibility that AI may destroy humanity, the least they could do is have a comparable amount of humility to entertain the possibility that human souls may be real and that they play an essential role in the continuation of evolution.
As a way of getting scientific skeptics to consider this possibility, the algorithm could be programmed to remind itself that there remains a modest-sized hole in the science’s reasoning. To wit, science still does not precisely know how biological life arose from “chemical soup.” (“Abiogenesis” and “panspermia” are two of leading scientific hypotheses today). Science also does not yet fully understand how the first single-cell organism gave “birth” to a multi-cellular organism. (There is the colonial theory, the cellularization theory and the symbiotic theory).
To the extent that these pesky evolutionary issues have not yet been definitively resolved from a scientific perspective, it would be wise for the creators of AI to not close the door on the notion that the soul – or something akin to an as-yet–discovered “transforming agent” – was behind these past evolutionary leaps and, thus, may be integral to future leaps.
The future fate of humanity may be at stake in getting AI’s algorithms just right. As such, the industry must ensure that every evolutionary path–however small or remote one may believe it to be–remains open and viable. To do otherwise is arrogant and may close the only hole through which humanity may escape.
Jack Uldrich is the futurist-in-residence at the Steger Leadership Center in Ely, MN and teaches strategic foresight at the College of Saint Benedict and Saint John’s University.