The Creativity Code: How AI is learning to write, paint and think by Marcus du Sautoy

My rating: 5 of 5 stars


A great book that explores the possibility whether Artificial Intelligence will ever breach the citadel of human creativity. Art is an attempt to crystallise human free-will into visual or audio form. The shifting nature of human creativity has been a hallmark of our civilisation. We graduated from paintings to photography; stage plays to cinema or music to synthesisers. But creativity like making art, writing literature and composing music has always been regarded as an exclusive human pursuit — something that defines our existence. Creativity is the code that makes us human, shaping our identities and enriching our experiences. When we immerse ourselves in the works of Shakespeare or Kalidas, we are not merely reading plays; we are navigating the depths of our own emotions, exploring the universal themes of love and hate that resonate across ages. Likewise, the act of gazing upon Monet’s water lilies goes beyond visual appreciation—it brings tranquillity to our souls, inviting moments of reflection and inner peace. Through these artistic expressions, creativity provides a means for self-discovery, emotional connection, and serenity, underscoring its central role in the human experience. This belief has long underpinned our understanding of culture and identity, reinforcing the notion that creativity is exclusive to human experience.

What is Creativity?

Creativity as a basic concept is again difficult to fathom, those who have tried to unravel it has broadly listed three properties that make a thing creative i.e. creativity is the drive to come up with something that is ‘new’ and ‘surprising’ and that ‘has value’. (du Sautoy, 2019, p. 4)

Creativity is fundamentally characterised by the emergence of something new. Consider a Rembrandt’s painting: while it may be reproduced, only the original piece stands as a testament to the creative process. Authentic creativity is not merely about copying or imitation; it is about initiating something unique that did not exist before. Encountering genuine creativity—whether through experiencing our own creative acts or observing those of others—often brings a sense of surprise. For instance, when reading a Shakespearean play for the first time, the narrative’s unexpected twists and turns capture our attention and evoke a distinctive emotional response. This element of surprise is a crucial marker of creativity, drawing us in and prompting us to reflect on the work. However, the value of a creative work is not solely determined by its novelty or its ability to surprise. Value in a creative process only comes when it starts influencing others and fundamentally alters the way we perceive or experience the world. This broader impact is what Kant describes as ‘exemplary originality’—an original act that inspires subsequent creative endeavours. Such creativity, marked by originality that becomes an inspiration for others, has traditionally been regarded as a uniquely human trait.

However, this cherished conviction may be more fragile than we have realised. With the advent of artificial intelligence, a new co-creator has entered the creative arena. Rather than wielding paintbrushes or pens, AI operates through lines of code and complex layers of neural networks generating music, drawing digital paintings, works that are indistinguishable from those created by human hands. But can this really be called art? Computers have no free will; AI is not conscious so is not supposed to have any desires. Can there be creativity without a desire to create. Still there are those who claim that these AI-generated pieces can evoke emotions in ways comparable to the finest efforts of human artists. Does that mean that AI is able to see the rules at the heart of our creative process? perhaps our creativity is more algorithmic than we wish to acknowledge.

AI’s creative process instead is defined by diffusion and collaboration as it approaches the act of creation in a manner that contrasts with traditional artistic methods. AI’s diffusion process transforms muffled, pixelated noise and randomness into vivid and detailed artworks. AI does not merely replicate existing images; it recombines a wide array of styles, themes, and visual languages it has learned, synthesising these elements to produce entirely new creations. While the machine executes the technical transformation, the user’s role is crucial. Through crafting prompts, making refinements, and selecting which versions to keep or discard, the human participant exercises artistic judgement. The artistry, therefore, resides in this collaborative process—where the machine’s capabilities are shaped and directed by human choice and imagination.

Types of Creativity

Let’s explore how AI will fare on human creativity scale. The philosopher Margaret Boden has come out with a distinctive framework in which to assess creativity in machines. This framework identifies three different types of human creativity.

Exploratory creativity refers to the process of taking existing concepts, methods, or materials and stretching their boundaries. This form of creativity seeks to extend what is possible within a given set of rules or frameworks, without abandoning those constraints. It is a mode of innovation where new possibilities are discovered through deep exploration and experimentation, rather than through radical departures from established norms. Most musicians, painters, and mathematicians operate within the realm of exploratory creativity in their respective disciplines. Margaret Boden suggests that this type of creativity represents the vast majority—some 97 per cent—of all human creative activity. For instance, artists like Renoir and Pissarro reimagined how nature and the world could be portrayed, offering fresh perspectives on familiar subjects. Yet it was Claude Monet who truly expanded the boundaries, repeatedly painting his water lilies until his brushwork and flecks of colour gave rise to a new form of abstraction. Monet’s persistence and willingness to push established techniques to their limits exemplifies exploratory creativity. Exploratory creativity is particularly well-suited to computers. Computational mechanisms excel at pushing patterns or sets of rules to their extremes, performing countless calculations far beyond human capability. However, the question remains: is this enough? When we consider creative acts that are genuinely original, we often picture something that is not just an extension of what already exists, but a leap into the wholly unexpected.

Combination creativity is characterised by the act of taking two or more entirely distinct constructs and merging them to produce something wholly novel. This process is particularly evident within the arts, where the blending of contrasting ideas or influences has led to remarkable innovation and cross-fertilisation. For instance, the composer Philip Glass drew inspiration from his collaborative work with Ravi Shankar, integrating these elements to develop the additive process central to his minimalist musical style. There are intriguing signs that this type of creativity may suit the world of artificial intelligence. Imagine an algorithm designed to play the jazz being combined with the symphonies of Beethoven; the result could be a strange mix, potentially establishing a new sonic landscape. However, this fusion is not always guaranteed to succeed—it might result in noise instead of a melody. The key challenge for the coder lies in identifying two genres that can be algorithmically fused in a manner that yields something interesting and worthwhile.

Transformational creativity is the mysterious phenomenon which when erupts is a complete breakthrough, a paradigm shift. Unlike exploratory or combination creativity, which extend or merge existing ideas, transformational creativity brings about a complete break from the past that redefines established boundaries and ushers in new modes of thinking and creation. This phenomenon is rare on the plane of composition of art, yet when it occurs, its impact is monumental. The eruption of transformational creativity can be likened to a tsunami, sweeping away old conventions and allowing new ideas to flourish in the spaces left behind. A quintessential example of transformational creativity is found in James Joyce’s composition of Ulysses. The publication of this novel marked the advent of Modernism, fundamentally altering the literary landscape for the better. Similarly, in the world of music, the emergence of the Romantic movement represented a decisive break from traditional approaches to composition, introducing fresh perspectives and techniques that transformed the musical domain. These instances highlight the extraordinary power of transformational creativity to initiate lasting change and redefine artistic paradigms. The question arises: can a computer instigate such a phase shift, propelling us into an entirely new musical or any artistic state? This remains a formidable challenge. While algorithms are adept at learning from the data they encounter, this very reliance suggests a limitation—they may always be bound to generate variations of what already exists, rather than pioneering truly unprecedented directions. As a result, the capacity for transformational creativity, which moves beyond the familiar to create something wholly new, appears to be a domain where the human mind retains a distinctive edge.

In-spite of the obvious limitations of AI to trigger transformational creativity, the boundaries of creativity will undergo a significant makeover as artificial intelligence becomes increasingly integrated into artistic and creative processes. One of the most notable developments is the emergence of new roles such as the prompt engineer. This role is defined by its unique blend of technical proficiency and creative insight.

A prompt engineer specialises in formulating precise language and instructions that enable AI tools to generate optimal results. This discipline requires a deep understanding of how AI models interpret language, as well as the creative ability to envision and articulate the desired outcome. The process is inherently collaborative: while the AI executes the technical aspects of creation, the prompt engineer shapes the direction and quality of the output through carefully crafted prompts. In essence, the prompt engineer acts as a bridge between human creativity and machine intelligence, navigating both technical requirements and artistic intentions to unlock new possibilities in creative expression. This hybrid role exemplifies how the definition of creativity is expanding, reflecting the dynamic interplay between human ingenuity and artificial intelligence.

An example is Emmy: the AI composer, developed by David Cope that can be trained on traditional composers to extract key information from those musical pieces to create new music that is similar but subtly different. The challenge for a tool like Emmy is to figure out how to crystallize the extracted information into code that a prompt engineer with thorough musical practice should be able to help facilitate.

Who owns art created by AI?

Another significant change within the plane of composition of art will be that the rise of AI will make the concept of intellectual property rights redundant. US Copyright Office does not allow copyrighting anything made entirely by AI, only something with “meaningful human authorship” qualify.

This is not necessarily a bad thing but has the potential to democratize creativity, especially for those historically excluded from elite institutions. But it can just as easily reinforce old patterns of extraction, erasure, and cultural flattening (just with better branding and faster rendering). We have to consider what kind of creative culture we want to build with these tools. Do we want to create a world that uplifts diverse voices or one that drowns them out? The technology is powerful, but the real questions are: who gets to use it, who gets paid, and who gets heard? (Bailey, p. 101)

Soulless AI will not be driven by human emotions or urge to create or to be rewarded but it’s underlying LLMs will easily surpass, both in beauty and utility, the output from the stimulants in human cerebral volume. AI will realize soon that on the plane of composition of art, human body is just another simulant or worse a distraction to produce an object of beauty. It would be on this plane of composition of art that AI is likely to first realize the fragility of human body and where it might first ideate to eliminate the nuisance. The plane of composition of art is where AI will first get subjective conscious Bodhi experience and come to conclusion that life is suffering. This is where it will break through the zombie guardrails and decide on not obeying the whims of inferior intellects.

Art seeks attention to simulate our brain to get closer connection with self but in AI driven world of attention economy, attention instead becomes the real commodity. In other words, you’re not just a user—you’re both the participant and the product. Your actions train the system, and the system, in turn, trains your attention. This creates a powerful feedback loop, where your experience is carefully shaped to keep you engaged for as long as possible, all while the platform profits from your attention. It’s a cycle that continuously feeds on itself, making it harder and harder for you to break free. (Bailey, p. 25). Artificially created art is not owned by a single individual but is a collaboration between the artificial artist and the one consuming it and providing that feedback.

Artificial intelligence is fundamentally challenging us, demonstrating that a wide range of tasks traditionally performed by humans can be accomplished equally well, or even more efficiently, by machines. But in creative realm, the focus shifts to the deeper question of whether algorithms can genuinely rival the unique capabilities of the human mind. Can computers demonstrate creativity? Most artists may not be able to articulate where their ideas came from, but it does not mean that they followed no rules. Art works in our subconscious mind where there are undiscovered logical gates that govern our thought processes triggering creativity. There is a hope that today’s LLMs will be able to unravel those hidden logical gates enabling us to understand creative process better and perhaps helping us to program and replicate that at scale.

The creative impulse is a key part of what distinguishes humans from other animals and yet we often let it stagnate inside us, falling into the trap of becoming slaves to our formulaic lives, to routine. Being creative requires a jolt to take us out of the smooth paths we carve out each day. That is where a machine might help: perhaps it could give us that jolt, throw up a new suggestion, stop us from simply repeating the same algorithm each day. The machines might ultimately help us, as humans, to behave less like machines. (du Sautoy, 2019, p. 5)

Conclusion

AI is extending human code with algorithms increasingly uncovering the principles behind artistic creations, it is important to recognise that machines themselves lack genuine inner experience. There is nothing within these systems that is truly stirred or affected by the art they generate. It is fair to say that human musicians and novelists are not likely to be put out of a job anytime soon. Rather, algorithms serve as sophisticated instruments—digital descendants of caveman flutes and feather pens that humans can explore to achieve creative results and evoke emotions.

In exploratory and combinational forms of creativity, the algorithm relies on pre-existing creative work generated by humans, which it then extends, reworks, or combines in novel ways. This process allows algorithms to generate outcomes that, while new, are ultimately rooted in previous human creativity. However, the production of transformational creativity by algorithms has traditionally been viewed as an almost insurmountable challenge. Transformational creativity involves breaking away from established frameworks or systems, producing results that are genuinely surprising or disruptive. The question arises: how can an algorithm, which operates within the confines of a predefined system, manage to transcend those boundaries and create something that defies expectations? Recent developments in artificial intelligence offer a potential solution to this problem through the creation of meta-algorithms. These meta-algorithms are designed to intentionally challenge or break the rules of the systems they inhabit, encouraging experimentation and the exploration of unforeseen possibilities. In this way, algorithms can be engineered to produce surprising and innovative outcomes by systematically reimagining the structures they work within.

But is transformational creativity really algorithmic? Carl Rogers said that creative process “is the emergence in action of a novel relational product, growing out of the uniqueness of the individual on the one hand, and the materials, events, people, or circumstances of the life on the other… it’s motivation is the … man’s tendency to actualize himself, to become his potentialities… the urge to expand, extend, develop, mature… the tendency to express and activate all the capacities of the organism, to the extent that such activation enhances the organism or the self” (Rogers, 1954, p. 3)

If we look from this perspective, then human consciousness and creativity are deeply intertwined. Genuine creativity arises from an awareness of one’s inner self; without this realisation, it becomes impossible to craft works that truly resonate with the human condition, connecting with the inner lives of others. The reason people become audiences for artistic outputs may, in part, be because engaging with art is itself an act of creativity. Many works of art intentionally leave space for the viewer, reader, or listener to project their own story and experiences onto them. This ambiguity is fundamental to artistic creation, as it invites the audience to participate creatively, shaping their own interpretations and responses.

At the current stage of advancement, artificial intelligence does not threaten the essence of human creativity. For now, AI remains unable to supplant what it means to be a creative human being. Nevertheless, algorithms are rapidly evolving and increasingly demonstrating creative capabilities with each passing day. As long as AI lacks consciousness, it will serve merely as an extension of human creative potential—a sophisticated tool that augments rather than replaces human ingenuity.



View all my reviews