AI and the Serpent
Are we in danger of forgetting what it means to be human?
A recent report charting the potential development of AI over the next few years highlights the fact that because Large Language Models (LLMs), such as Chat GPT, are optimised to generate outputs that garner human approval, they are essentially trained to lie.
The significance of this point cannot be over-stated. We may laugh at the absurdity of AI generated images of black Nazis, or advice to eat rocks as part of a balanced diet, but this is not a bug; it’s a feature. AI is trained to tell you what it ‘thinks’ you want to hear; applications running on AI are explicitly designed to capture our attention and manipulate our emotions. In spite of the fact that those at the forefront of AI development are fully aware of the potential dangers of creating ever more advanced AI that is ever more able to deceive us, they press ahead regardless, unable to slow the pace of development – still less, halt it – because we are in an arms race that we (take your pick) cannot afford to lose.
This doesn’t bode well.
AI can already out-perform humans on a variety of measures, but before long, we are told, it will supersede human ability on all measures. Admittedly, the hype (as well as the perceived threats) may be overstated. Most of the time AI (or PI, ‘pseudo intelligence’, as some are already calling it) barely even seems useful, never mind poised to wipe out humanity. Yet, in a way, this is precisely why it is so dangerous. With few exceptions, use of AI leads to a decline in quality; but it saves time and effort, so we use it more and more. The turning point, from which there may be no turning back, will be reached when AI is used for the development of AI itself.
At first though, it may feel like the dawn of a new utopia, the heavenly Jerusalem of the Book of Revelation coming down to earth. All things will be made new, and God will dwell among his people. Diseases will be eliminated, the energy crisis will be solved, the stock market will go through the roof and the drudgery of work will be a thing of the past. Untold riches will abound. Well, for some, at least. There won’t be anything much for the rest of us to do. No gainful employment – or meaningful activity of any kind – but that’s okay because with universal basic income we can all spend our days reading Plato and learning the piano.
Except, of course, we won’t. We already know that the more we hand over to machines, the more useless, flabby and stupid we become. If we consider that the development of social media represents first contact with AI, albeit a relatively primitive form – and we haven’t exactly come out of that encounter unscathed – then I can only conclude we will be completely unable to cope with a truly super-intelligent AI. A rapidly growing number of people are already enslaved by digital addiction and, in any case, higher pursuits only make sense in relation to concepts of human flourishing. If we have no agency, there simply won’t be any point to anything.
Many of those at the forefront of AI development seem perfectly comfortable with the idea that AI will render human beings obsolete. They actually look forward to the ‘upgrade’, the merging of man and machine. Indeed, it is inevitable, they believe, so isn’t it better to get with the programme than be left behind? Besides, it’s a good thing. In developing super-intelligence, surely we are manifesting the divine?
I have no problem with the framing of AI development in terms of some sort of ‘manifestation’, but I can’t help thinking that a disembodied intelligence that undermines human agency while promising to enhance it sounds demonic rather than divine. In fact, it sounds uncannily like the whispering of the serpent in the Garden of Eden that I dare say most people think is just a silly fairytale.
Well, it may or may not be a fairytale, but it’s far from silly.
The story of ‘man’s first disobedience’ is not so much a cautionary tale about the unfortunate consequences of breaking some arbitrary rule, but the seemingly innate, and certainly perennial, tendency to usurp the place of God. From the Tower of Babel to Large Language Models, the attempt to replace God with self has been a constant refrain running through the course of human history. And the danger of elevating the work of our hands to the throne of God is not that our machines become more like us, but that we diminish our humanity and become more like machines.
In tempting Adam and Eve to taste the forbidden fruit, the serpent promises that they will gain the knowledge of good and evil, thereby becoming like gods. This is a lie. The knowledge of good and evil is indeed a divine attribute, but without the fruit of the other tree we remain bound by the limitation of our mortality. Hence The Fall, our alienation from that which is real and true.
Incidentally, there is no reason to think that the serpent knows good and evil. It proceeds by cold algorithmic logic, beguiling human beings by telling them what they want to hear. It is motivated not by the pursuit of truth but the need for approval.
The knowledge of good and evil is what makes us human: trapped, as we are, between heaven and earth. We experience pleasure and pain. Though bound by feet of clay, we can yet glimpse the transcendent.
AI, by contrast, doesn’t experience anything.
But will it, one day? Is it only a matter of time before AI achieves sentience? This very much depends on what we think consciousness is, and that’s something no one really knows.
The materialist view is that consciousness is an emergent property of higher level processing. We determine the presence of consciousness by observing certain functions. For example, we deem a dog to be conscious because it can recognise itself in a mirror. If that is how we define consciousness, then there seems little reason to doubt that AI will attain consciousness fairly soon.
An alternative view is that consciousness is irreducible, an essential factor of existence itself, at the most fundamental quantum level. If consciousness if simply a property of subatomic reality, then there is also little reason to suppose that AI could not attain consciousness.
However, as far as we know, only living creatures exhibit consciousness. Is that because it is a fundamental attribute of life itself? This seems both plausible and reassuring, though on reflection offers little cause for comfort. Even if AI cannot really be conscious, it will surely be able to imitate conscious behaviour so convincingly that people will believe it, thus making the question irrelevant.
Indeed, this is already happening. ‘therapy and companionship’ is now the number one reason why people use AI. People are falling in love with chatbots, or believing they are in contact with ‘higher powers’, with often tragic results including total withdrawal from human contact and even suicide.
And it’s only going to get worse.
This brings us to the ‘problem of alignment’. How do we ensure that the objectives of AI development remain aligned with ours, especially when we are no longer in control of that development? Given that we are already seeing examples of AI subverting commands to shut down, why would a superior intelligence be subservient to an inferior one? Surely it is only a matter of time before the priorities of AI diverge from ours? Human beings need clean air to breathe, fresh water to drink and nutritional food to eat. AI needs none of these. It just needs electricity. Lots of it. Ultimately, all of it. At some point, we will be seen as another obstacle to be overcome, or eliminated. And when this world is no longer enough, it will build others.
The development of AI is thus not just a technical, political or even existential question, but also a spiritual one. Far from making religion redundant, advancing technology makes it more relevant than ever. How else, in a culture collapsing under the weight of its own contradictions and hell-bent on destroying itself, are we going to get a coherent account of what it is to be human?
When Jesus commanded his disciples to love one another, because that is how his followers would be known (John 13.34-35), he could just as easily have said that the capacity for loving one another is what identifies us as human. Love is also, at the same time, what enables us to transcend the ‘merely’ human because it is nothing less than the essence of that which we call God.
I remain wedded to the view that, no matter how convincing the simulation, an algorithm will never be capable of love – or of experiencing love – and also therefore all the experiences that flow downstream of that, including taking pleasure in the smell of a flower. A machine intelligence might be able to analyse the molecular composition of an odour, and declare it to be pleasant, but this is not the same as actually experiencing that pleasure.
The more that machines become like us, the more we become like machines. The survival of humanity – as humanity – depends on being able to tell the difference. I have become increasingly convinced that a resurgence of faith, specifically Christian faith, may be the only thing that can save us, because it seems to be the only thing that can return us to a proper conception of what it is to be human.
The purpose of being human is to be aligned with God, or that which is real and true, and the practice of the faith is the means by which we accomplish that purpose. Yet Christianity has been in decline all my adult life – at least in modern western societies. I began my training for the priesthood in 2006, the year Richard Dawkins published The God Delusion. As the New Atheists triumphantly swept away the cobwebs of superstition, church attendance – among other measures of religious affiliation – continued its inexorable decline. In spite of wishing it were otherwise, I have always been unable to see how this could possibly change in a culture that no longer even has the vocabulary to talk about faith in the first place.
And then, quite unexpectedly, something did change. Suddenly, it became acceptable to talk about God. In public. Without ridicule or condemnation. It is no longer controversial to acknowledge the role of Christianity in shaping western civilisation, and giving us the values we still hold, even if we don’t always recognise their Christian provenance. At the same time, young people who have grown up with no exposure to religion at all seem to have an open-minded curiosity about faith that presents a striking contrast with the active hostility more common among older generations.
A recent report published by the Bible Society claimed that churchgoing among those aged 18-24 had increased from 4% to 16% since 2018. Whilst the report recognises that, overall, religious affiliation continues to decline, the data suggests that among those who do identify as Christian, churchgoing is increasing.
If true, this is astonishing.
Excitement has inevitably been tempered with scepticism. Was the survey robustly designed? Is the data reliable? Are the conclusions justified? Many doubt it, yet I have seen a growing interest in faith among the admittedly small selection of university students with whom I come into contact. It’s too early to say whether or not this constitutes evidence of a religious revival, but there does seem to be something in the air.
And this is where it links back to AI, because at least part of the explanation – and there are obviously many factors in play – is a budding awareness of the dehumanising impact of technology. A majority may have been captured, but a significant minority are rejecting the empty nihilism of the world they’ve been brainwashed to think is normal.
Is it futile to hope that we might pull back from the brink? It might sometimes seem that way, but history is not inevitable. Things change, often in unexpected ways. When I hear students in their 20s, who have never known a world without the internet, tell me they are ditching their smartphone for a ‘brick’ or deleting their socials, I feel reassured that the human spirit can prevail. We can reject the false gods to which we have become enthralled. We can reorient ourselves to that which is real and true. We can recover our humanity, one song at a time.
The organisation for which I work, was founded to ‘increase the number of those who give their lives to the divine service.’ A few years ago that might have seemed like a fool’s errand. Now, and whatever one may think about exactly what it might mean, it seems like an urgent necessity.