Human Creativity and AI Creativity

By Anton Suslonov · Interdisciplinary essay · Published July 26, 2026

What you are about to read is quite unusual. It is the quintessence of an interdisciplinary approach.

We will try to compare human creativity and the creativity of artificial intelligence, drawing on ideas and methods from three fields of knowledge that usually seem entirely separate.

Three fields of knowledge

The Austrian school of economics (Menger, Böhm-Bawerk, Mises, Hayek) gives us the understanding that a person is their values, as well as their tastes and ideas of the beautiful.
The Moscow school of psychology (Vygotsky and Luria) helps us understand how inner language forms and how discursive thinking develops. In an unexpected way, this is rather close to how neural networks operate on tokens.
Mathematics and the theory of machine learning - knowledge of how artificial intelligence works on the inside: neural networks, large language models, and neural generation of images and sound.

What do we know about the human brain?

We more or less know:

what working elements the brain is made of;
how those elements interact;
how those elements store information.

Direct experiments confirm:

that memory is states of cells and ensembles of brain cells;
that the visual cortex works somewhat like a convolutional neural network;
that thinking is linked to the activation of neurons.

Experiments allow us, to some extent, to read, decode, and reproduce signals the brain receives and transmits.

We know that the brain’s operation resembles that of a neural network and is also based on a threshold principle: a natural neuron “opens” when the sum of signals at its inputs exceeds a threshold level, just as an artificial one does. A natural neuron is, of course, far more complex than the virtual model.

A risky digression

Suppose the person is undecided: they want both and cannot tell which they want more.

…and then a cosmic ray arrives - a truly random event - and breaks an enzyme molecule somewhere in the liver.

The nervous system gets a signal that “something is wrong with digestion,” and the desire for meat disappears.

The person is given an extra nudge: “Come on, decide already where we’re going!”

In a neuron - imagine it is one such “music neuron”; a simplification, of course - the value rises above threshold. It opens, and the person decides to go listen to music.

How does the human brain create works?

A person has a living brain structure, its current state, a body and its state - what we can call a “personality.”

Diagram of the human creative process

And how does AI create its works?

Diagram of a generative model

Diagram of the AI creative process: idea, large language model, autoregressive cycle, and physical embodiment

What is the difference?

Now let us place these two pictures side by side and compare how they differ.

As is easy to notice, in AI’s creative process there is:

no human cycle of deliberation (there is a generation cycle, or a mechanical outer cycle);
no personality;
no personal experience and personal knowledge of the physical world;
no values;
no personal taste and personal ideas of the beautiful;

There are only memories of the world, layered one upon another.

That is not yet creativity. If a person draws together with artificial intelligence, creativity does not exist without the evaluating and deliberating element - and that element is the human. AI is a generator of ideas and images that still need judgment and thought.

Can AI become “like a human”?

The answer is, in principle, yes - if we add a cycle of deliberation to the LLM.

That is called AGI, Artificial General Intelligence, whose approach Anthropic’s marketers regularly announce, and which has not been reached yet because LLM architectures are not suited to a deliberation cycle. My forecast: it will happen within 10–20 years.

Architecture alone is not enough, though. Experience is needed, and values are needed. Where would they come from? AI has no family; personal and evolutionary survival look entirely different for AI; it is unlikely that AI can be raised the way a human is raised.

On that optimistic note, we will stop.

Enter for creators