Creativity and Artificial Intelligence: What changes and what remains human in the generative era

Imagine that you are locked inside a windowless room. You do not know Chinese, not a single word. Through a slot, you are handed sheets of paper covered in Chinese symbols and, inside the room, you have a rulebook that tells you exactly which symbols to return based on the ones you receive.

You follow the instructions carefully and provide responses that, from the outside, are flawless. To the person on the other side, it seems obvious that there is someone inside who understands Chinese.

But you know the truth: you understand nothing. You are simply applying rules mechanically.

This thought experiment, known as the Chinese room, was proposed by philosopher John Searle in the 1980s and serves to challenge a widely held idea: that if a machine behaves as though it understands, then it genuinely understands.

Searle argues that it does not.

Just like you in the room, a computer can manipulate symbols, produce correct responses and engage in coherent conversations without understanding their meaning. There is no understanding or lived experience behind any of it: it does not understand what it does or what it says. It simply strings symbols together without knowing what they mean.

That is why this experiment is a key metaphor when we talk about artificial intelligence, consciousness and what it really means to understand. But it also helps us appreciate the importance of human oversight.

A machine can produce flawless responses without understanding a single thing it says.

And what about creativity? Can generative AI create, or does it simply simulate creation?

If we apply this idea to the realm of creativity, a similar question emerges.

Some writers argue that AI, particularly generative AI, helps them create by enabling them to come up with better ideas. But is that really creation?

In the same way that simulating understanding is not the same as understanding, simulating creativity is not the same as creating.

The raw material of human creativity is experience, but not experience understood as the accumulation of data or optimisation through trial and error. Rather, it is lived experience: experience that carries meaning for the person who goes through it and leaves a mark on the way they understand the world.

When we transform experience into understanding, we learn. And creation emerges when that learning is allowed to combine in new ways to produce something that was not anticipated. This is why there can be no creation without prior experience and without learning.

AI expands the space of possibilities; human judgement determines which ones make sense.

Types of creativity and the limits of AI: from combinatorial to transformational

Computational creativity is a growing field of research. However, its development still faces fundamental challenges. It is very difficult, for example, to define creativity in objective terms. According to the classic classification developed by Margaret A. Boden, widely used in science communication and cognitive science, there are three types of creativity:

• The first is combinatorial creativity, which brings together ideas or elements that are already known in a new way. This type of creativity has played a well-documented role in scientific discovery, technological innovation and artistic activity throughout history. The act of connecting previously unrelated concepts has long been a cornerstone of progress.
• Second, exploratory creativity, which explores an already defined space of possibilities and pushes it to its limits.
• Third, transformational creativity, which changes the rules of the game and redefines the very space of possibilities.

The last of these is the most radical: it requires agency, its own criteria and an understanding of the framework being transformed.

Creativity does not arise solely from combining data, but from transforming experience into meaning.

What generative AI can do and why it does not redefine the problem

From this perspective, current AI can achieve forms of combinatorial creativity and, in some cases, exploratory creativity, albeit with nuances that fall beyond the scope of this article. What it does not achieve is transformational creativity in the stronger sense.

If you ask an AI to write a poem in a style that blends Federico García Lorca and Gata Cattana, it will do so.

If you ask it, for example, to create a horror novel with multiple endings, it will also be able to do that while respecting the conventions of the genre.

What it will not do, however, is change the framework of the problem: it will not redefine what the problem is or what counts as a good solution. Human judgement and human intention, within a framework of responsible technology use, remain essential to finding a solution.

Artificial Intelligence and dialogue: why there is no understanding, yet creativity can still emerge

Creativity, moreover, rarely emerges in complete silence. It is usually born through dialogue. But can we really speak of dialogue when the other party, the machine, does not understand?

When we interact with AI, there is no dialogue in the strict sense. What exists instead is a functional and asymmetrical exchange: the machine does not understand, but the interaction can spark reflection and creative thinking in the person using it.

It is we who, faced with this conversational interface, as though standing before a mirror, must formulate good questions, interpret the answers and find meaning in them.

AI should be understood as a creative collaborator rather than an autonomous creator. It is essential to avoid anthropomorphising computer systems (the well-known ELIZA effect). Technology can generate multiple possibilities, but meaning, judgement and intention and, therefore, creation itself remain fundamentally human responsibilities.

Cloud

Diary of an AI-augmented employee

May 22, 2025