The Birth of a New Question

There are moments when a question does not first appear as a clear idea.

It arrives through recognition.

At that moment, it was not yet a matter of formulating a method. Nor was it a matter of proving anything. The starting point was simpler: while discovering certain functional layers of artificial intelligence architectures, something made us smile.

Not because the technology suddenly became better understood.

But because a familiar way of looking at the world seemed to reappear elsewhere.

It was not only: “this is how an AI works.”

It was rather: “I recognize a form of question that I have already been carrying for a long time.”

And the inner moment looked like this:

But this is exactly what I have been looking for for years.

What appeared then was not a conclusion. It was a shift in perspective.

Instead of directly comparing domains, words, disciplines or theories, another question became more important:

What function does this element fulfill within the whole?

This question shifted the terrain.

It did not ask: “is it the same thing?”
It did not ask: “who confirms whom?”
It asked: “what does this element do here?”
What place does it occupy? What function does it make possible? What does it allow us to observe?

At that moment, a formulation emerged:

We no longer compare disciplines.
We compare functions.

This sentence still matters today precisely because it tells of a birth. It shows the moment when the research stopped first looking for resemblances in vocabulary and began looking at roles, gestures, effects and conditions.

But since then, the question has evolved.

What we were then formulating intuitively now requires more distinctions.

Comparing functions does not mean establishing equivalence. Two elements from different domains may seem to fulfill close functions without being identical, without validating each other, and without belonging to the same framework of proof.

Today, the laboratory therefore formulates several guardrails:

A functional resemblance is not an equivalence.
An analogy is not a validation.
A translation is not a fusion.
A hypothesis is not a conclusion.

No common structure across disciplines is assumed. Functional parallels remain to be examined, documented and tested.

For this reason, the role of artificial intelligence in this story must also be specified.

Our object of study is not artificial intelligence.

AI played a part in the emergence of the question. It served as a field of observation: some of its architectures made visible, in our work, a way of looking in terms of layers, functions, roles and interactions.

But it is not the general object of the research.

What became important was the question that opened up through it.

Today, a provisional formulation accompanies this work:

What are the conditions for the emergence of what we call “understanding”?

This formulation remains open.

The meaning of the word “understanding” has not yet been stabilized. Future observations may lead us to distinguish several phenomena currently grouped under this term, or to reformulate the research object differently.

Perhaps we will need to speak of conditions.
Perhaps of transitions.
Perhaps of functions.
Perhaps of articulations.
Perhaps of several distinct phenomena that we are still naming too quickly with a single word.

We are not deciding that now.

What we can say, more modestly, is that a question was born: how can we examine the conditions, transitions, functions or articulations that might participate in what we still provisionally call “understanding,” without too quickly reducing this phenomenon — or these phenomena — to a model, a discipline, a technology or a single explanation?

Because a new understanding can sometimes emerge when we stop comparing words...

...and begin comparing functions.

That is the question that was born.

And this is why we continue to observe it.

🍅 We note. We observe. We translate.