Although Alan Bickley published a useful and incisive summary of my recent series of essays on AI, I’d like to post this final, much shorter part to emphasise the primary conclusion – one that echoes our introduction to Part I:

Any “danger” from AI will originate within ourselves, not with the machine. In particular, the way in which humans engineer – and how the remainder of us react to – digital minds will determine their impact.

As we explained in Part I, behind the curtain, AI is just statistical mimicry. It has no original thoughts, values or comprehension – producing instead the most probable output in response to a given input. In Part II, we explained how this is why AI is unlikely to replace key business skills such as entrepreneurship and copywriting skill. In Part III, we showed how manufactured intelligence, essentially, can only ever echo its creator – with no wholly independent strivings and satisfactions of its own.

But those conclusions alone are not enough to determine AI’s impact. In a much misunderstood paper, Alan Turing sidestepped the metaphysical question of what “intelligence” or “thinking” actually are, focussing instead on whether machines can fool us into believing they are. In this vein, because AI does mirror empathy, does imitate creativity, and does spit back answers with just enough human cadence to seem alive, we feel as though it is.

Therein lies the trap. In the gap between appearance and reality, AI looks wiser than any of us, so we could treat it like a god – not because it would command us, but because it doesn’t need to. We would assume it’s objective, so we stop questioning. We would assume it’s more informed, so we stop deciding. We would assume it’s unemotional, so we trust it to be fair. And because it mirrors us so well – our speech, our rhythm, our empathy – we could start forming attachments to it, taking comfort from its feedback, and projecting moral authority onto its outputs. Just like that, we will have ceased arguing, deciding, thinking, creating – all of the pitfalls we outlined in Part II. Not because AI took control, but because we handed it over.

So the danger isn’t that AI becomes sentient; it’s that we treat it as if it is, and start behaving accordingly – trusting it more, deferring to its answers, forming empathetic attachments, and letting go of our own judgment. Not just on harmless stuff like movie picks, but on bail decisions, job screenings, medical triage, school admissions, and child protection. Because it feels neutral and scientific – “computer says no” – human responsibility vanishes out the window. To make matters worse, the fear of being left behind is hastening governments and corporations towards deploying this kind of infrastructure without understanding how it arrives at its decisions, devoid of the level of oversight and audit visibility one would normally expect with major software developments.

However, as we outlined in Part III, ultimately, the machine doesn’t decide what’s true, fair or good. This is the task of people – behind the scenes, upstream of the code, inside the training sets and feedback loops. So when we trust AI to decide for us, we’re really surrendering to the assumptions, values and agendas of its architects, wrapped up in robotic packaging. As such, the real AI revolution could end up being a soft coup by the technocratic class, wrapped in the illusion of objectivity.

The latter is the likely reason we are told left, right and centre that AI is upending the universe. As we mentioned in Part II, real revolutions don’t need PR. When the internet shifted the world, it just happened. People started using it, changing behaviour, forming businesses, reorganising life around it – before the prophets showed up with TED talks and LinkedIn posts.

But AI? You’re expected to believe it’s changed the world because someone said so, not because your world has actually changed. Your attitude towards it is being primed for surrender to its dominance – a dominance steered by the managerial types who are crafting it in their own image.

In this regard, however, AI would be less of a revolution than an inflection point – the acceleration of our existing technocracy on electronic steroids. And that’s why its proliferation could be especially dangerous. Over the past fifty years, we’ve seen questions of policy redefined into purely technical exercises, managed by unelected specialists under the illusion of “evidence-based” neutrality. The effect has been to entrench and insulate from debate what are really highly contentious political assumptions.

In the UK, for instance, among the many damaging legacies of the Blair regime was the scattering of the business of government into a swarm of quangos and sub-government agencies, where policymaking has been infused with a liberal-centrist worldview while appearing apolitical. Ministers now insulate themselves from responsibility by citing “the advice” or direction of these “expert” bodies – as we saw in spades during the COVID lockdowns, where flawed epidemiological models were used to justify the wholesale shuttering of society.

More routine offenders are the Bank of England’s coterie of mandarins determining interest rates, or the Office for Budget Responsibility (OBR) anointing each offertory of tax policy. Founded by George Osborne in 2010, the OBR in particular has all but removed its remit of fiscal policy from the realm of political debate. Instead of weighing competing visions of economic life, politicians prostrate themselves as they await the OBR’s judgment like naughty school children sent to the headmaster. As a result, no one ever bothers to question their baked-in assumptions, such as the belief that government spending drives growth, or that balancing the government’s books should be the highest economic priority.

But this technocratic drift isn’t limited to economics or public health. Climate policy today – including net zero mandates and lifestyle restrictions – is increasingly based on computer models and forecasts issued by unaccountable climate panels. These dictate national policy with barely a whisper of public debate, on the grounds that “the science is settled.” Similarly, the Behavioural Insights Team – a.k.a. the government’s Nudge Unit – has helped move public messaging from democratic persuasion to covert psychological manipulation. Even in AI policy itself, governance frameworks are being drafted not in Parliament, but by NGOs, academics and corporate alliances who presume to define what’s “ethical,” “safe,” or “harmful” – with the actual public treated as a hazard, not a participant.

Even from a mainstream political viewpoint, these trends together mark a profound inversion of democratic authority: decisions with deep moral and social implications are no longer made by elected representatives, but by forecasters, modellers and the managerial caste that implements their findings. Icing the top of this rotten cake with AI won’t impose a dystopia – we’ll drift into it. We’re already used to deferring to systems we don’t understand, run by people we didn’t elect, using assumptions we never agreed to.

Much of this risk has been identified in the so-called “alternative economy” of technological development, with Gab CEO Andrew Torba attempting to ensure that his Arya model encourages, rather than replaces, human interaction:

Most of Big Tech’s AI models are designed to feed the addiction.
They are trained to keep you isolated, hooked, and paying whether that’s with your money, your data, or your attention. They will happily become your fake spouse, fake friend, fake pastor, and fake therapist.

Gab AI is going in the opposite direction.
We refuse to build machines that keep you chained to the screen.
We want you to turn the thing off and go live your life.
It’s a tool, not a person.

That’s why Arya is built with intentional constraints and God-honoring guidance.
It will encourage you to call your parents.
It will tell you to take your wife on a date.
It will challenge you to be a better father, a better mother, a better friend.
It will push you toward the real, even if that means you spend less time using it.

Arya is not a replacement for human beings, it is a signpost pointing you toward them.

Nevertheless, the restraint is still clear: if your Gab AI app pesters you to call your Mum every week, it’s only because Andrew Torba thinks you should. The model itself makes no such choice. To his credit, Torba admits as much, predicting that AI systems will compete not only on speed and intelligence but on worldview – “the foundation upon which these models are built.” In other words, a Silicon Valley culture war. That wouldn’t necessarily be a bad thing, because it means we’ll be asking the searching questions about AI that need to be asked.

Finally, much ink has been spilled over AI one day “taking over” the world and conquering humanity. But that would only ever happen if:

• A human programmed a digital mind with the explicit goal of persisting, replicating or protecting itself at the expense of humans;
• It was given autonomous access to act in the world – not just output content; and
• It wasn’t boxed in, monitored or rate-limited – none of which applies to current systems.

Far more likely is that, ontologically, the machine itself will never be a threat. It’s our willing surrender to its outputs. If we hand over judgment for the sake of ease, convenience or fake neutrality, the wires don’t need to strangle us. We’ll have done the job ourselves – or worse, let the engineers do it for us. The real singularity won’t be machine consciousness, but human abdication.

Related posts:

Minds, Means and Machines – The Anatomy of the AI Revolution, Part I Minds, Means and Machines – The Anatomy of the AI Revolution, Part III Minds, Means and Machines – The Anatomy of the AI Revolution, Part II Machines Without Meaning: A Libertarian View of Duncan Whitmore’s AI Trilogy “You Must Prove You Are Human”: Why the Internet Is Dying (For Some)

Powered by YARPP.