Reflections from the PKM Summit 2026 — Utrecht, March 2026
In his presentation at the third PKM Summit in Utrecht, Jorge Arango dismantled a metaphor that should have been questioned long ago: the “second brain.” It was about time. Few expressions have seduced so many while saying so little — or rather, while saying something plainly wrong. A note-taking tool is not a brain. A graph of linked notes is not a thought. And the fact that thousands of practitioners adopted this slogan without batting an eye says less about the metaphor than about our appetite for intellectual shortcuts.
One can be forgiving, though. PKM — Personal Knowledge Management — is a nascent discipline. Every nascent discipline goes through its phase of excessive slogans and poorly calibrated promises. The emergence of tools like Obsidian, Logseq or Notion marks a genuine turning point: the shift from a world where we searched for information to a world where we must survive its saturation. New practices had to be invented. New myths, inevitably, had to be invented too. The “second brain” was one of them.
But what happened in Utrecht this year calls for a more uncomfortable reflection.
AI Everywhere, Thinking Nowhere
While it was not mentioned at all during the first edition of the PKM Summit, and only briefly evoked in 2025, this year, conference after conference, artificial intelligence was omnipresent. In the demonstrations, in the audience’s questions, among PKM trainers as well as framework developers and plugin builders. With very few exceptions, every speaker approached AI as a natural extension of knowledge management tools.
At the time, it seemed obvious to me. Natural. Inevitable, even.
I chose to write this article a few days after the summit — because stepping back has become a luxury that is almost unaffordable in a world where everything demands an immediate reaction. With a bit of distance, the picture becomes considerably more disturbing.
We started by externalising our memory. That was defensible. The original reasoning behind PKM held up: our brain is made for thinking, not for accumulating. Taking information out of our heads and entrusting it to an organised system meant freeing up cognitive capacity for what truly matters — thinking, analysing, creating new connections.
Then we added artificial intelligence. And now, it is no longer memory that we are externalising. It is the combinatory work. The ability to connect ideas, to detect hidden patterns, to produce syntheses. In other words: what looks remarkably like thinking itself.
The Suspicious Praise of Zero Friction
One word kept coming back insistently in the corridors and on the stages: friction. Or rather, its elimination. The ideal tool would be the one that removes all friction between a thought and its capture, between a question and its answer, between a need and its solution.
But friction is not a flaw in the system. Friction is effort. It is precisely the intellectual effort we must produce if we want to think with any degree of rigour. It is the effort that forces us to reformulate, to compare, to doubt, to start over. It is also the effort through which we retain what we have learned. To suppress friction is to suppress the very condition of learning.
When a “frictionless” tool summarises an article for me, suggests relevant links and generates a synthesis before I have had time to read the first line — at what point do I think?
The Flight Forward
What I observe, with yet another step back, looks less like progress than a flight forward. We are entrusting the management of our personal knowledge to something entirely impersonal. The paradox should be staring us in the face, but technological enthusiasm has this remarkable property of making invisible what is right before our eyes.
Let us add the problem of confidentiality, which is far from anecdotal. Entrusting one’s knowledge network — notes, reflections, intuitions, doubts — to systems hosted in the cloud means offering up to machines what constitutes us intellectually. And these machines are not neutral: they collect, they learn, they feed on what we give them. With each exchange, the machine learns. We, less and less.
The Announced Atrophy
There is a simple principle that everyone can verify in their own body and their own experience: any function left unused atrophies. A muscle that is no longer solicited wastes away. A skill that is no longer practised withers. A capacity for thinking that is systematically delegated eventually weakens.
If PKM systems augmented by artificial intelligence become prostheses for our basic cognitive functions — memory, association, synthesis, judgement — then the question at stake is no longer merely one of efficiency. It is a question of identity.
And this is where the loop closes with a cruel irony. One can extend Arango’s reasoning to its logical conclusion: not only is there no “second brain” — but if we continue on this trajectory, it is our own brain that we are rendering superfluous. In favour of an artificial brain endowed with limitless memorisation capacity and an output that resembles genuine thought more closely with each passing day — even though nobody, for now, seriously believes that it is.
The question is no longer whether the tool is useful. It is. The question is at what point we stop being those who think assisted by machines, and become the assistants of a machine that thinks in our place.
Or, to put it more bluntly: are we becoming the prostheses of our own prosthesis?
I have been modelling the workings of the first brain for thirty years. The only one. The real one. Because a PKM that ignores how we think, learn and decide is just another storage tool. A PKM that takes it into account becomes an amplifier of intelligence — not its substitute.