For Deleuze and Guattari, especially in What Is Philosophy? (1991), a concept is not a general idea, a definition, a representation, or a label for things already given. A concept is something created in order to make a problem thinkable.
Their famous formula is:
Philosophy is the creation of concepts.
That sentence is much more radical than it first sounds.
A concept, for Deleuze, has several features.
First, a concept is a multiplicity. It is never a simple unit. Every concept is composed of several heterogeneous components held together in a certain consistency. Take Descartes’ cogito: “I think, therefore I am” is not merely one proposition. It coordinates doubt, thinking, existence, the first person, certainty, and so on. A concept is therefore something like a constellation rather than a box.
Second, the components are not arranged hierarchically like genus and species. They form what Deleuze and Guattari call a zone of neighborhood or indiscernibility. The elements touch and transform one another. A concept has an internal topology.
So we might write:
concept ≠ category containing instances
but rather:
concept = multiplicity + relations + consistency.
Third, concepts are created in response to problems.
This is already central in Difference and Repetition. A genuine problem is not simply a question waiting for the correct answer. The problem has its own structure; it generates possible solutions.
Thus:
concepts do not simply answer questions; they articulate problems.
For Deleuze, this is enormously important. A bad philosopher often gives excellent answers to badly posed problems. A great philosopher changes the problem itself.
Nietzsche does not merely answer:
What is morality?
He creates the concepts of will to power, ressentiment, genealogy, and thereby changes what the problem “morality” means.
Bergson creates duration.
Spinoza creates immanence, affect, conatus in a particular configuration.
Leibniz creates the monad and, in Deleuze’s reading, the fold.
The concept therefore belongs to a philosophical event.
And this leads to perhaps the most Deleuzian point:
A concept is an event, not an essence.
Traditional philosophy often asks:
What is X?
and seeks the essence underlying many instances.
Deleuze is much more interested in:
What is happening here?
What relations are forming?
What is becoming possible?
So the concept does not primarily represent an object.
It captures—or better, constructs—a pure event.
The concept of “becoming,” for example, does not describe a class of things called becomings. It allows us to think transformation without assuming that transformation is merely movement between two fixed identities.
Hence:
concept ≠ representation of what something is
but:
concept = construction of what can happen / become.
Another crucial point is that concepts exist only upon a plane of immanence.
This is one of the hardest but most beautiful parts of What Is Philosophy?
A philosopher does not merely invent isolated concepts. Concepts operate upon an implicit image of thought—a field determining what counts as thinking, being, relation, difference, truth, etc.
Deleuze and Guattari call this the plane of immanence.
Thus:
plane of immanence → concepts
Spinoza’s concepts belong to one plane.
Descartes’ concepts belong to another.
Nietzsche’s to another.
Concepts therefore cannot simply be removed from one philosophy and inserted unchanged into another. They acquire meaning through their relations on a plane.
There is another companion to the concept: the conceptual persona (personnage conceptuel).
Descartes has the idiot—the thinker who doubts everything.
Nietzsche has Zarathustra.
Socrates has the friend/interlocutor.
These are not literary decorations. They are figures through which thought moves.
So Deleuze and Guattari propose something like a triad:
plane of immanence — the field on which thought takes place
conceptual personae — who thinks
concepts — what thought creates.
This is also why Deleuze insists that concepts have dates, signatures, and histories.
There is no timeless dictionary of philosophical concepts.
“Substance” in Aristotle, Spinoza, and Whitehead is not the same thing.
“Difference” before and after Deleuze is not quite the same concept.
A philosopher signs a concept almost like an artist signs a work.
That does not mean concepts are arbitrary. A good concept must possess what they call consistency. Its heterogeneous components must hold together.
This gives us another useful formula:
A concept has consistency without having identity.
That is extremely Deleuzian.
It can remain a multiplicity without being reduced to One.
And this explains something else: concepts are not propositions.
Science works mainly through functions and propositions describing states of affairs.
Art creates percepts and affects.
Philosophy creates concepts.
So in What Is Philosophy? they make a famous tripartite distinction:
Philosophy → concepts
Science → functions / functives
Art → percepts and affects
None is superior. They are three different ways of confronting chaos.
This phrase is worth emphasizing:
Thought confronts chaos.
Philosophy does not simply organize already settled reality. It cuts into chaos and constructs consistency.
So I would compress the Deleuzian concept into this:
A concept is a created multiplicity of heterogeneous components that gives consistency to a problem on a plane of immanence.
But there is an even shorter version that I suspect may be closer to what interests you:
A concept is a machine for making something thinkable that could not previously be thought in quite that way.
And therefore concepts are not primarily for classification.
They are for production.
That is why Deleuze is so different from a dictionary-style philosophy.
A dictionary asks:
What does this concept mean?
Deleuze asks:
What does this concept do?
What connections does it permit?
What old distinctions does it break?
What new problem does it construct?
What new form of life or thought becomes possible through it?
This is why concepts such as rhizome, deterritorialization, body without organs, becoming-animal, assemblage, difference, repetition, and virtual can initially feel strange.
They were never meant merely to name pre-existing objects.
They were designed to force thought to move differently.
And here I think there is a particularly useful connection to your recent discussion of fragments and AI.
A Deleuzian concept is not simply a synthesis of fragments into greater coherence.
It is often produced when previously distant fragments are made to enter a new relation:
fragment A × fragment B → new problem → concept.
That is why Deleuzian philosophy is closer to construction than to system-building.
AI might be extremely good at organizing existing concepts.
But the genuinely Deleuzian moment may occur when someone says:
“What if these two things that supposedly do not belong together actually belong to the same problem?”
That is already the birth of a concept.