Introducton

The term 'Technological Singularity' is one of the most discussed, but also one of the least precisely defined, concepts in the AI conversation. Ray Kurzweil's 2005 book The Singularity is Near did a lot to popularise the term, but the notion of a 'Technological Singularity' is a lot older and has come to mean slightly different things to different groups of people.

Eliezer Yudkowsky has written a great review (2007) about the different definitions of the 'Technological Singularity', as the term has been used by different writers. He distinguishes three different but overlapping schools of thought around the concept

The first is the way the term is used in Kurzweil's book, referring to a moment where the speed of technological change will go so fast that it becomes impossible for humans to keep up with the changing world. The second way in which this term is used (Vernor Vinge's event horizon view) is largely similar, but emphasizes how the future becomes fundamentally incomprehensible and unpredictable to humans beyond this moment.

Often this Singularity moment is related to the onset of Artificial General Intelligence (AGI). This is originally due to the mathematician and statistician Irving John Good, who argued that once an AI can improve its own cognitive abilities, you get a recursive feedback loop that rapidly produces superintelligence far beyond anything human.

The term 'Technological Singularity' is compelling in popular discussions around AI, because it sounds like a clear definition, and gives a name to a real fear: that technology might accelerate beyond what any one person can understand or control, and that one day humans will no longer be able to understand what is going on.

That is exactly why the term gets so much attention. It sounds like a sharp threshold, a moment where history splits into a well-defined before and an after. With technology and society operating like a black box beyond our comprehension after the Singularity.

It seems a relevant and compelling idea as we are seeing AI Agent emerge everywhere. But as an analytical concept, I would like to argue here that it is not a very workable definition when we try to discuss in more detail what is going on in the world. As a result, discussions often become semantic debates about when the Singularity might till take place, instead of concrete discussions about what is actually changing in the world.

In many ways the rapid progress in AI is not so different from how earlier technological advances that caused the pace of technological progress to go so fast that it grew beyond the understanding of every individual human.

There is a famous thought experiment (originally from Leonard Read's 1958 essay "I, Pencil") that makes this point beautifully: there is no single human being on the planet who understands every detail of how to produce something even as simple as an ordinary a pencil.

The wood comes from specific types of trees that need to be logged with specific equipment. The graphite is mined and processed using chemical engineering. The metal ferrule, the rubber eraser, the yellow paint, the glue, each component involves its own global supply chain of specialised knowledge that no one individual fully comprehends.

And yet pencils get produced by the billions. The system works, not because any one person understands it all, but because millions of people each understand their own piece in it, and the whole system is coordinated through markets, institutions, and shared protocols.

In the same way, there is not a single human capable of explaining everything an AI-driven economy does. But that doesn't mean the system is out of control. It means the system is complex. And we have been living with complex systems we no longer fully understand for centuries, or even millenia. The Singularity, in that sense, is less of a cliff edge. The moment when artificial intelligence begins improving itself is more a continuation of a trend we've been on for a very long time.

When did we pass the Singularity?

If we try to look for a clearly markable milestone like passing th Turing test, or passing the Singularity in the 21th century, then I'd argue the key milestones look something like this:

If we consider all the data storage, processing, and transfer carried out over the internet, we might argue that we already passed the moment where 'the total amount of artificial cognitive power surpasses human intelligence'.

We passed the Turing test (or at least all practical version of it) around 2022, when ChatGPT launched and millions of people suddenly found themselves having conversations with a machine that felt, for all practical purposes, like very close like talking to a knowledgeable human. Yes, you could still trip it up. Yes, it hallucinated. But the threshold of "can a machine convince an average person it's intelligent in casual conversation?" was crossed.

Somewhere around 2025 Agentic AI systems started to meaningfully contributing to AI research: Writing code that accelerated the development of the next generation of AI, helping with the implementation of new architectures and optimisations. When the thing you're building helps you build the next version of itself better, you've entered a feedback loop. But this is no different from every previous technological revolution where a new technology helped us to build an improved next version of it. Yet we arguebly passed the moment already where both 'AI is helping to improve itself' AND 'AI surpasses human intelligence'.

Did the world end? Did everything become incomprehensible overnight? No. There was no clearly identifiable before and after. The Singularity didn't arrive as a dramatic event. It is still arriving as a gradual shift. Which, looking back, is exactly how every previous technological revolution has felt to the people living through it at the time.

Some might say there is another better definition of the Singularity.

I would like to demonstrate in the next posts that throughout modern history we have already had 'collective superintelligent' entities (every organisation being an example) that were formed out of networks of humans. As we are now adding AI systems into these organisations, and allow them to take care of an increasing share of the cognitive work, there will again be a smooth transition, with no clearly definable before and after to distinguish between the days wehre AI systems 40% vs 60% of the cognitive workload of any process in such an organisation.

I will do my best to cearly distinguish between what is actually changing, and what is still exactly the same as it has always been as we are now living in what we might truely call a Post-Singularity Society. What an exiting time to be alive!