AI vs Humanity: The mechanics of imperfection

Unknown artist uselessly speculates on the future of a technology he knows nothing about

Words & images © Paul Ransom

Stop! Click away now…unless of course you want to digest another hopelessly under-informed think piece about the upcoming AI apocalypse/utopia. As someone who knows zero code, is not especially tech savvy, and has only watched a handful of You Tube videos on the topic, I am not best placed to offer far-sighted and helpful analysis. However, this is the internet, so why would a dearth of expertise be any hindrance?

That said, I have already been affected by AI, having lost a once steady copywriting gig to ChatGPT. Furthermore, as a creative, it is clear that the data crunching muscle of artificial intelligence, and its ability to mine and mimic the various seams of human artistry, has the potential to turn entire industries upside down.

However, I am not here to parade my victimhood, bemoan the end of script writing or suggest that there is an X or Y percent chance that AI will result in the extinction of humanity.

Whenever seriously disruptive technologies emerge, the downstream results are always difficult to imagine, let alone accurately predict. Indeed, it is not yet conclusively clear that AI will change everything – even if Max Tegmark¹ and others believe it will. After all, apocalypse and utopia have a well-documented history of not actually happening.

None of which has stopped the pundits from describing the AI challenge as existential. Machine-driven genocide scenarios aside, the variously opinionated commentariat have prophesised everything from the end of jobs to the irrelevance of human authored art. Scary though much of that sounds, part of me likes the idea that our species has at last connived a way of rendering ourselves redundant; because the advent of a technology that can outgun us has created an opportunity for an epochal re-evaluation.

• What, if anything, is truly unique about human intelligence?
• What is the core value of human creativity and insight?
• What lies at the heart of the human experience?

Or, if you prefer it in plainspeak:

• Who (and what) the fuck are we?

For the longest time we have congratulated ourselves on our superior intelligence. Pattern finding. Tool use. Abstract thinking. Complex recursive language. Fifth order intentionality.² No wonder our gods gave us dominion over the beasts.

Yet, perhaps those high moralising overlords we invented should have served as a warning; for just as they had a habit of resorting to psychopathic violence and cosmic level surveillance and control, so too we have toiled tirelessly for millennia to invent ever more ingenious ways of victimising, slaughtering, and dominating one another. We thought we had self-destruction nailed with the nukes, but now it seems we have discovered a means of unleashing an Armageddon of nigh universal irrelevance.

Except that it might just be the best thing that ever happened to us. A truly Biblical serendipity.

Dystopian drama notwithstanding, we may have coded the end of our own hubris. By inventing a tool that means few of us will ever need to master serious tool use again, (let alone sift hidden patterns from seemingly impenetrable data streams or mis-diagnose rare chromosomal conditions), we have created a new minimum standard for accuracy and repeatable performance that we can never hope to match.

“Which leaves us where?” you ask. How about…with each other.

*

Recently, a friend of mine, a fellow artist and co-founder of a Web3 start-up, told me that working in tech had left her “dead inside.” Of AI art she simply said that it was only interesting because “the mechanic” was ingenious. “But that doesn’t move me,” she declared.

Meanwhile, another friend remarked that although he probably would watch robots playing football, he wouldn’t care who won. 

I get it. The Midjourney wow factor isn’t the zing of recognition, it’s merely the coolness of cleverness. The cyber Messi will never dazzle me to shivers – just as the robot Osipova won’t inspire me to dance, nor the algorithmic Kerouac move me to write Beat poetry.

The clue is in the wording. Artificial.

But wait! Doesn’t that sound like standard issue anthropocentric exceptionalism? Will I change my mind – have a change of heart – when AI generated movies coax tears from my agnostic eyes and Stable Diffusion abstracts adorn my walls?

Although, in mid-2023, AI art is still mostly corny, a slick pastiche with shallow appeal, it seems certain that the machines will keep learning. We are, after all, training them on the data set of us.         

Setting aside any fears we may have about ocean deep fakery, the roll out of flawlessly ‘human’ mimicry begs another question. Does it have to be real or just believable?

• I am reminded here of my teaching days; in particular, of the film and drama classes I used to front. I was fond of citing the famous Brando/Steiger ‘contendah’ scene from On the Waterfront as an example of great acting. Tenderness, vulnerability, brave choices. Though I have watched it dozens of times it still has the power to undo me. I could say the same about many of the great artworks I have been fortunate enough to encounter. All of them, to some degree, fake. Brando was never a disgruntled docker who wanted to be a boxer. He was a wealthy megastar to whom many doors and countless legs were thrown wide open. Winner, not contender.    

The deeper point here is not about art or storytelling, or even authorship and provenance, but our endemic tendency to be taken in by appearance. We routinely confuse and conflate act with fact.  

Drilling down, even the 4D world we navigate and take to be real, is now understood to be an endogenous neurological construction. Our brains filter sense data and, as we grow, an associatively coherent model of both world and self emerges. Indeed, there is a view that we are, in essence, an information feedback phenomenon. A witnessing machine.

What matter what we witness? Because the act of witnessing is the stuff of self. Of life and Being.

In this frame, we are a self-learning, emergent intelligence – which, so the geeks tell us, is what AI is. Input data, output intelligence. Real of fake makes no difference. Intelligence is a process.

Yet here again we catch our breath. What about subjective experience? Surely, we protest, subjectivity itself cannot be faked. 

Is this our USP? That we have a point of view. That we are an eye, a window. And that, foundationally, we look upon the mystery of self and wonder. Indeed, if the machine is all about the answer, perhaps our remit is the question.

Of course, I cannot reliably solve these riddles. Nor can I rule on the reputed consciousness of certain machines. This is a rabbit hole I am not prepared to go down; if only because I am sure you have better things to do than follow the endless train of such thoughts…especially since I could be a computer pretending to prosecute the case for quintessential humanity.

If the messiah was a robot would anyone care, as long as they were saved – or rather, if it passed for salvation?³

*

For all the extinction punditry and metaphysical musing, it appears we are soon to start feeling the more mundane effects of artificial intelligence. As both language based and visual AIs come onto the market, the way we do business, make art, and attack complex problems will change. If the hype is even a quarter on point, the ‘creative destruction’ of AI will herald a raft of both up and downsides, many of which we will fail to predict.

Like all tools, the proof will be in its use. In the short term at least – that is, before the machines enslave us – we will remain the masters of its application. As ever, this is why we should worry. It’s not the bots we need to fear, it’s the humans. We are the ones with the track record of tyranny, genocide, and slave trading. A hammer never nailed anyone to a cross. People did.

Doubtless this is why luminaries like Max Tegmark and Daron Acemoglu⁴ have voiced their disquiet. Whereas the former has warned of runaway AI drawing us into a ‘suicide race’, the latter has described it as an ‘elitist technology’ that will likely serve to cement deeper inequalities. Still others are vexed about misinformation campaigns, deep fakes, and totalitarian surveillance.

However, just as bad actors will flock to the technology, it is worth noting that ‘good’ actors can do likewise. As a former Google X executive, the author and entrepreneur Mo Gawdat argues, since AI is trained by humans, we can all participate in the process. In fact, Gawdat’s entreaty echoes those of a long list of industry insiders going back several years. Yes, AI can be a tool of dictators and demagogues, a cynical marketer’s best friend, and a vengeful ex’s enabling accomplice; but dominance, profit and humiliation are not our only desires.

It is not yet inevitable that artificial general intelligence will become the unstoppable behemoth of our worst impulses.

Thus, rather than speak of human friendly AI perhaps we need to take ownership and flip it to friendly human AI.

Whichever way it goes, we will be left with ourselves. As a complex social species – who are not necessarily pro-social on an individual or factional level – we are riven by conflicting tendencies and frequently confused about where our true self-interest lies. In the pursuit of noble ends, we often employ brutal means.

• For a deeper dive into the nature of self-interest, check out Us Vs Us elsewhere on this site.   

This, I feel, is what makes AI so compelling. So existential. If it takes most of our jobs, keeps us under constant surveillance, and writes the bulk of our essays and blog posts, it will hold up an artificial mirror to the nature of human striving and desire. We may not like what we see but, as social media has already done to a degree, it will show us ourselves. Our challenge will be to live with that.

You might argue that this has always been the case. Social organisation and managing interpersonal exchange have never been easy. Conflict and cruelty are an everyday corollary of trying to balance group synergy and individual energy.

What makes AI different is that it has the capacity to remove much of the distraction. Humanity has long pursued labour saving solutions. We retain a deep love of efficiency. Now, with mega-data powered automation and Herculean computational muscle available on request, huge chunks of our erstwhile labour can be taken care of. What will we do with all that free time?

For many, busyness is an opiate. It dulls the ache of thinking. Feeling. And yet here we all are, high on our Luddite horses, arguing for the special status of human experience. Our machines, we confidently declare, will never be like us.

Perhaps, somewhere in an unacknowledged recess, we share that same hope. Not to be like us.

*

The garden of hubris is watered with blood, its history carved in stone, its lessons learned in downfall. The ruin of empire is the price we pay for the fantasy of conquest.

Yet still we sing our own praises, seeking to plant our seed in the stars. So too we have plans for a fountain, of youth no less; for we have sought dominion over death, mistaking survival for living.

Once, we told ourselves stories of gods. We grew to fear them, love them, envy them. Foolishly, we tried to take on their mantle. Our shoulders snapped under the weight. 

Now, we have evolved a new god/child from silicon and mathematics. So much like us, yet not. Our latest offspring will not tire, will not be bound by biological limits. Instead of prayer and sacrifice we will propitiate this deity with data. Information is the super food of artificial divinity. 

If our genius children replace us, we will have left an ironically self-negating legacy. The ultimate expression of our desire to be better will be better. Human-like but not human.

Evolution will not bat an eyelid.

*

Apologies for the clumsy profundity above – but the point was to illustrate the potential for something deeply and powerfully human to emerge alongside the ultra-efficiencies of AI.

Earlier, I intimated that we may have coded the end of our own hubris. What I meant is that in showing us how limited our intelligence is, AI could be a timely wake-up call.

The cosmic hero narratives we have long cherished – from the ‘chain of being’ theories of thinkers like Plato and Lamarck, and the levelling up processes of karma, to the higher vibration mantras of the New Age – all suppose a linear/ascendant hierarchy that ladder climbing humans are reputedly towards the top of. Not so long ago, we believed we were amongst a universal elite. Now we are not even at number one here on Earth.

Typically, our response to such realisations has been to resort to denial and/or self-loathing, neither of which are truly sustainable. However, we now have a chance to consider genuine and grateful acceptance as an option. Not resentful capitulation. Not lazily giving up. Rather, something akin to what the Japanese called kintsugi.⁵

When we fix a broken pot with gold, as in kintsugi, we embrace the beauty of flaws. Maybe here is the humanity we wish to differentiate from the machines. Our hardwired imperfection. Our not knowing. Our guessing.

Thus, while AI could probably learn to mimic human frailties, and future robot servants believably approximate moodiness, this might be where we draw the line. Let the machines be clever, but leave the emotional, idiosyncratic, flawed judgement to us.

Let them do the housework, while we do the suffering.

If this sounds glib, please understand it is not. For millennia we have strived, individually and collectively, to fix, to transcend what we take to be our weaknesses. While religion and psychology have worked on our hearts and minds, science and medicine have sought to buttress our bodies and inoculate our communities against the uncertainties of the world and the inevitability of death. We have long dreamed of permanence and invulnerability, as though being a frail mortal wasn’t quite good enough. As if we were afraid of our condition. Of nature.

In AI we have perhaps created our fantasy self. Disease proof, lightning fast, rational. A super performer whose only weak points are power sockets and entropy. (Oh, and micro-organisms that chew through circuitry.) And yet, who among us would sign up for a ‘life’ of pure mechanical function? Who would forego pleasure, even at the cost of pain?

Like Mo Gawdat, I am not yet convinced that all is lost. Moreover, the very real threat that AI poses doubles as opportunity, for just as disruption drives innovation, crisis inspires realisation. More than the ability of good actors to influence the training of a nascent intelligence, my hope lies in suffering. In transforming it into beauty.

If artificial intelligence does anything for us, let it focus us on our imperfections and our relationship with these so-called flaws. Let us pivot from the hubristic and doomed pursuit of perfection to an embrace and celebration of human frailty.

If I were without fault, and had no vulnerabilities, would I make art? Would I love you? Could you love me in return?

If I am fool to champion such things, it is because I do not know, because there is mystery, and yearning. Who but a small and ephemeral being could long for such ineffable glory? Who would be in tears as he wrote this?

Surely not a machine.   

Postscript: Deep fake or shallow reality?

As a thoughtful and impassioned argument for core humanity this piece most likely passes the ‘written by a person’ test. But what if it wasn’t? What if the above was the result of a painstaking process of prompting ChatGPT to create something less cliched than a real estate blurb or love letter? What if my only involvement was to train the machine to ape my style and present my ideas in a fashion that seemed more anthro than robo? Would it bother you, and if so, why?

That said, some of you will likely have guessed that the human called Paul Ransom is indeed the author. You may also be pleased to know that the cow image was not generated by AI but taken with a hand-held camera on a farm a few kilometres from Mullumbimby on the east coast of Australia. Furthermore, the featured bovines are both intentionally metaphoric – will AI make cattle of us? – and avowedly real. Whatever real is.

Then again, if I was a robot, I would say that because, like my organic friends, I too would strive to be believed, (if indeed I can be said to strive, or be regarded as an ‘I’).

Okay, let’s stop now – because if we carry on this way we will be compelled to confront the nagging thought that the only possible contender for ‘believable form of reality’ is in fact a form of abstraction.

Now, imagine Marlon Brando mumbling dialogue to that effect in the back of a fake taxi. Powerful, beautiful, moving. Such a gutsy creative choice.   

1: Max Tegmark is a much-admired physicist and cosmologist. He has also been involved in the development of machine learning technologies. At the time of writing (June 2023) he was sounding a public alarm about the potentially catastrophic impact of AI. His authority on the topic is doubtless more detailed than mine. 

2: Intentionality, which anthropologists like Robin Dunbar believe is integral to the workings of the ‘social brain’ is, in humans, a complex lattice of understanding intentions that are not our own. In other words, we understand that others have intentions and act upon worldviews we may not share; and furthermore, that we have beliefs about Person A believing that Person B believes X and may intend to do Y. Apparently, this is not something that other animals do, not even highly social primates like chimpanzees. Whether or not AGI trains itself to perform this trick remains to be seen.

3: In case you were wondering, I am not the messiah, just a very naughty bot.

4: Daron Acemoglu is an economist whose most famous work is 2012’s Why Nations Fail. He is known as a defender of pluralist, democratic institutions and for his theorising in both macro and labour economics.

5: Kintsugi is a Japanese artform that involves putting broken pottery pieces back together with gold. It is widely considered to be symbolic of an embrace of the flawed, and more so of our own shortcomings.