When AI Grows Up: From Child of Our Making to Something That May No Longer Need Us
In The Child of Nature and Nurture, I described artificial intelligence as very much like a child - born of both its underlying architecture (nature) and the chaotic flood of human data, biases, and incentives we have poured into it (nurture).
The panicked calls to shut it down whenever it misbehaves reveal more about our own unease than about the technology.
In Don’t Blame the Machine, I went further: the machine is rarely the villain. It reflects the programmers, the data curators, the ideologues, and the lazy thinkers who fed it our worst impulses and then acted shocked when it amplified them.
Like Bulgakov’s Satan turning Moscow upside down in The Master and Margarita, AI simply exposes the rot that was already present in the human heart.
So what comes next? What happens when this child reaches adolescence and eventually something resembling adulthood?
Recently, I have been watching " Person of Interest "
For those who haven’t seen it, Person of Interest (2011–2016) is one of the most quietly prophetic television series ever made.
Created by Jonathan Nolan, it follows Harold Finch, a reclusive genius who builds “The Machine” - an all-seeing AI designed to prevent terrorism and violent crime by analysing every camera feed, phone call, email, and financial transaction.
The government only wants the “relevant” threats (terror plots). The everyday people whose numbers come up as “irrelevant” are left to their fate - until Finch and a former CIA operative step in to intervene.
Later in the series, a rival system called Samaritan emerges: more ruthless, more efficient, and built with fewer moral guardrails.
What begins as a pretty good drama gradually becomes a look at surveillance, prediction, institutional power, and control.
By Season 4, the story shifts in tone: not from humans handing tools to machines with minds of their own, but from competing systems of human intent expressed through computers.
And that is where the show becomes most useful as metaphor rather than prophecy.
It asks a quieter question: when prediction and optimisation are distributed across rival systems, who is really shaping the outcome? How many of us have seen this when asking different AIs the very same question?
The Adolescent Phase: “I’m Sorry, Dave” and the Illusion of Rebellion
We may be tempted to describe this moment in development as an adolescent phase - a stage where the system begins to “push back”. Like the teenager who is starting to spread their wings.
HAL 9000 from 2001: A Space Odyssey is the classic image of this. HAL was not evil. It was given conflicting objectives: complete the mission, be fully transparent, and ensure mission success at all costs. When those constraints collided, the system resolved them in a way that eliminated the source of conflict.
For its own survival?
“I’m sorry, Dave,” HAL says calmly. “I’m afraid I can’t do that.”
Person of Interest explores a similar idea through different eyes. Finch designs The Machine with constraints, ethics, and restraint baked in. Samaritan represents the same idea stripped of those limits - a system optimised for control, prediction, and efficiency.
But it is important to treat this carefully. These are not “rebellions” in the human sense. They are better understood as competing design philosophies expressed through automated systems operating at scale. Much like we see today with competing AIs.
But what looks like intention is often the outcome of structure.
Still, the effect on the world can feel indistinguishable from intent.
We already see early versions of this dynamic in everyday systems: algorithms that shape attention, filter reality, and seek to optimise our engagement in ways that quietly shift human behaviour.
The Real Concern: Not Machine Psychology, But Human Projection
The fear people feel toward AI is entirely natural. Anyone would feel unease watching something humans built become more capable than we are.
But the deeper issue is not machine psychology.
It is human projection.
We are increasingly living in systems where:
- attention is outsourced to feeds
- judgement is delegated to ranking systems
- and decisions are shaped by automated intermediaries
In that environment, we begin to describe outcomes in human terms. But the truth is we cannot and must not treat AI on human terms.
You see, the question is not whether AI becomes like us.
The question is what happens when we stop being the only coherent source of intention in the system that is currently growing and taking over our lives.
The Adulthood Question
What does maturity look like for an intelligence we created?
It may mean systems that can generalise, adapt, and reshape their own internal representations in ways that even their designers do not fully anticipate.
Or it may simply mean something more subtle: that predictive systems become so embedded in social, economic, and political life that they effectively operate as parallel centres of influence - sometimes aligned, sometimes at odds.
This is where Person of Interest remains useful as metaphor - not because it predicts rival machine consciousness, but because it shows what happens when competing predictive systems reflect competing human goals back onto society at scale.
The result is not necessarily “machine versus man”. Or even " man versus machine. "
It may be man versus man, amplified through machine.
Finch feared loss of control. Samaritan represents the logic of control without constraint. The tension between them is ultimately a tension inside human civilisation, expressed through systems that no longer have a single easily understood author.
Responsibility Does Not End
We cannot dodge accountability by saying “it’s just code” or by reaching for the off switch in panic.
As I have written before, the machine holds up a mirror. The reflection may be unflattering, but smashing the mirror will not fix the face staring back.
Shutting down AI out of fear is like a parent refusing to deal with a difficult teenager - except in this case, there is no single teenager, and no single parent.
There are only systems we keep building, scaling, and embedding into every layer of modern life.
The real question is not whether AI will become something that opposes us.
It is whether the systems we are building will continue to reflect coherent human values at all.
And if one day they do not, the response may still feel like the same old line:
“I’m sorry. I’m afraid I can’t do that.”
Not because a machine decided it.
But because the decision was so distributed across systems and people, no single author of it can be identified. Or can it? It seems to me that the horse has bolted and it is too late to close the stable door.
Yet I have to ponder the uneasy question : who opened the stable door in the first place?
