The Sorcerer’s Apprentice

We are about to embark on a ten-blog series of thought pieces I’ve entitled “Loving the AI-lien” (apologies to David Bowie for the bastardisation). The title is only half a joke. AI has landed like an unannounced visitor from elsewhere, and this series is about the project management profession learning to work with AI rather than fear it. Before the series proper begins, it is worth asking the question underpinning any discussion on AI:

“Was AI built with a purpose to answer some defined question, or simply because it had become possible, and possible things get built?”

The honest answer is less tidy than either proposition might indulge, for I would argue that the answer is “both”. That is not a fence-sit. It is the shape of the evidence. AI began with a specific, stated intent. It was then carried far past its founders’ original purpose by other forces: curiosity, prestige, military need, and commercial fear. Three lenses bring the picture into focus: the historical, the philosophical and the economic. We’ll take each in turn.

The history is unusually clean about AI’s origins. In the summer of 1956 a small group gathered at Dartmouth College for a research project. The group consisted of John McCarthy, Marvin Minsky, Nathaniel Rochester and Claude Shannon. In the proposal that convened them, they coined the very term “artificial intelligence.” The seed of AI lay in the hypothesis they would test, that “every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.” This was not an accident falling out of some other invention, but a question deliberately posed: “Can human thinking be formalised and mechanised?” However, a seed and its weather are different things.

What the founders of AI could not supply was the money, and the money, when it arrived, had ideas of its own. Long before anyone had a product to sell, Cold War governments funded the young field of AI, both heavily and early. Those first outcomes included code-breaking, machine translation, and command-and-control. In the USA, much of the first two decades of progress rode on defence budgets through the defence research agency we now call DARPA. This meant the field’s direction was shaped at least as much by who paid, as by what the founders were curious about. The seed was purposeful; the growth was steered by its sponsors. This “Origin-versus-Momentum” distinction is worth holding on to because it is the whole argument writ small. Was AI destined, or did people keep choosing to build it?

The philosopher reframes this as an older quarrel: choice versus determinism. The determinist case is seductive. Some things get made simply because they have become makeable, and the sheer appeal of a hard problem is enough motive in, and of itself. Robert Oppenheimer, of all people, gave the instinct its motto — that “when you see something that is technically sweet, you go ahead and do it,” and argue about what it is for, once it works. Applied to a machine that might be made to think, the problem is irresistible. It advances whether or not anyone has settled on why. This is the Sorcerer’s Apprentice in a lab coat. Enchant the broom first, find the purpose later, build the stop button ‘when’ and ‘if’ you need it.

Against that sits a sharper objection, and not to be mistaken for optimism. Critics of the determinist story argue that AI is unavoidable but not inevitable, and the two words are not synonyms. Ubiquity is not Destiny. Technologies look inevitable only in the rear-view mirror. Up close they are a bundle of choices and accidents. Why this matters, is not academic. If AI is simply destiny, we stop asking what it is for and carry on adopting it. “It just happened” quietly launders away the people who decide, and the accountability and responsibility that entails. “Inevitable” is a convenient thing to believe if you happen to be a fortune teller selling the future. A rejection of “inevitability” means that every step on the AI path was, and remains, a human choice.

Economics explains the pace and it does so in a single word, “rivalry”. Once AI looked useful, building it stopped feeling optional. Nations framed it as a contest for primacy. Vladimir Putin’s much-quoted line that “the leader in this sphere will become the ruler of the world” is only the bluntest version. Defence money followed the rhetoric on the settled view that AI would “define the next generation of warfare.” United States defence spending on AI, big data and cloud computing rose from roughly US$5.6 billion in 2011 to about US$7.4 billion by 2016, and by 2025 estimates of annual United States military spending on AI run into the low tens of billions of dollars.

Private capital dwarfs even that. For example, in January 2025 the Stargate project was announced. It is a private venture to build large-scale AI data centres in the United States, with an investment figure of up to US$500 billion over four years. By comparison, China’s private AI investment reached about US$9.3 billion in 2024, against roughly US$109 billion in the United States (Stanford AI Index, 2025). The US–China rivalry now pulls hundreds of billions, public and private, along behind it. What this produces is a race dynamic, and a race supplies a purpose of its own, unlovely but powerful: build it before the other side does. That is neither pure curiosity nor a single agreed goal. It is fear, dressed as strategy.

So, was AI built with purpose or was it inevitable? The same evidence licenses several honest verdicts. You may weigh the founding evidence and conclude AI was built with a purpose in that defining Dartmouth question, then a sequence of concrete goals from translation to diagnosis. You may weight how the breakthroughs actually arrived, that deep learning waiting patiently for enough data and computing power, and conclude it advanced because it could. That the purposes were discovered after the fact to fit the capability. You may throw up your hands and call it inevitable. That the individual motives are lost in the collective drift, recognising that this inevitability should be held lightly, as it conveniently absolves everyone inside its bubble of responsibility from any negative outcomes.

I would argue that the position that best fits the whole arc is the least dramatic. AI is both of purposeful origin and a product of momentum-driven growth. That is, a deliberate research question in 1956, propelled by Cold War funding, the thrill of the problem and a rivalry driven global economic race, far beyond anything its authors had in mind, resulted in something new, glistening with potential and possibility. This would account for the clean founding goal and the modern sense that the thing now advances on its own. This is precisely the apprentice’s predicament. His trouble was never that the magic had no purpose. It was that the purpose was overtaken by the momentum.

Which is exactly why a series on AI and project management ought to begin here, with motive rather than mechanics. If the lesson of the last seventy years is that capability tends to arrive with a will of its own, looking for a purpose afterwards, then the habit a project professional should cultivate when handed such a capable new tool is the reverse one: fix the purpose first, and admit only the capability that serves that purpose.

That is the ethos of SSLM. A framework is a deliberate act against “just because we could”. It is the decision to “build the minimal that meets the need”. It is the capability to move the water and put the broom back in the cupboard when the floor is dry. Technology will not supply that restraint; it never has. We professionals will. The rest of the series is about how we may do this in the projects we are responsible for. It helps to know before we start, that the machine was built the way most powerful things are: on purpose, and then beyond.