Digital Economy Dispatch #197 -- Exploring the Creative Use of AI: How to Understand Creativity in Delivering AI-at-Scale

Digital Economy Dispatch #197 -- Exploring the Creative Use of AI: How to Understand Creativity in Delivering AI-at-Scale
18th August 2024

One of the most important debates today about the increasing use of AI concerns whether AI enhances or replaces human creativity. In particular, should we view most AI systems as little more that very clever compute-intensive pattern matching? Or are AI systems capable of demonstrating human-like actions that offer new insights and experiences? It's a fascinating topic that challenges our understanding of both artificial intelligence and human cognition. Yet, more than that, our perspective on these key questions forms a critical piece of the puzzle when considering where, how, and when to deploy AI-at-Scale in organization.

There are many ways to explore this area. But recently I have been drawn to the work of Margaret Bowden. She is a research professor of cognitive science at the University of Sussex, where she has worked since 1965. She is a pioneer in the field of artificial intelligence and cognitive science, and has made significant contributions to our understanding of human creativity and its potential replication in machines.

Boden's work on creativity in AI is particularly noteworthy. She has been instrumental in bridging the gap between cognitive science, computer science, and philosophy, providing a framework for understanding and potentially replicating creative processes in artificial systems. Boden defines creativity as "the ability to come up with ideas or artefacts that are new, surprising, and valuable." This definition provides a useful framework for examining how AI systems might demonstrate creative capabilities. Based on this, she has introduced the three types of creativity – combinatorial, exploratory, and transformational – which have become foundational concepts in discussions about both human and artificial creativity.

Let's take a closer look into Boden's three types of creativity and see how AI measures up.

Combinatorial Creativity

Combinatorial creativity involves combining familiar ideas in unfamiliar ways. It's about making unexpected connections between existing concepts or elements. This type of creativity is perhaps where AI shines the brightest.

AI systems, particularly those based on machine learning and neural networks, excel at finding patterns and relationships in vast amounts of data. They can combine elements from diverse sources in ways that humans might never think of. For example, AI-generated art often blends styles and techniques from various artists and periods, creating unique visual compositions.

Many examples of this form of creativity can be seen today. For example, we see AI systems produce music that fuses different genres, generating songs that sound both familiar and entirely new. In the realm of product design, AI can combine features from multiple existing products to suggest innovative solutions. This ability to rapidly explore combinations of ideas makes AI a powerful tool for enhancing human creativity in fields ranging from marketing to scientific research.

Exploratory Creativity

Exploratory creativity involves working within an established conceptual space or style and pushing its boundaries. It's about discovering new possibilities within a given framework.

AI systems have shown remarkable abilities in this area, particularly in games and problem-solving. The most famous example is probably AlphaGo, which not only mastered the complex game of Go but also developed strategies that human experts described as beautiful and creative.

Typically, today we see AI systems explore the space of possible solutions in fields like drug discovery and materials science. These systems can generate and evaluate millions of potential compounds, identifying promising candidates that human researchers might have overlooked.

In the arts, AI can explore the stylistic space of a particular genre or artist, creating new works that are consistent with the established rules but push them in unexpected directions. For instance, I am sure we’ve all played with AI tool such as Udio and Suno to generate music in the style of Bach or Beatles that can sound both authentic and novel.

Transformational Creativity

Transformational creativity is the most profound type, involving changing the rules of the conceptual space itself. It's about breaking established paradigms and creating entirely new ways of thinking or doing.

This is where the question of AI creativity becomes most challenging and controversial. Can AI systems truly transform conceptual spaces in the way that, humans do? Is there any AI equivalent to intuition, inspiration, or surprise?

While AI has not yet demonstrated clear examples of transformational creativity, I believe we're seeing glimpses of this potential. Some AI systems are capable of generating hypotheses in scientific domains, proposing new theories that challenge existing paradigms. In the arts, we're seeing AI-generated works that defy traditional categorization, potentially creating new genres or forms of expression. In some of these cases, we can’t explain how AI works or how it knows things that no one told it.

However, it's important to note that these transformations are often the result of human interpretation and framing of AI outputs. The AI itself may not "understand" that it's breaking rules or creating new paradigms. This raises philosophical questions about the nature of creativity and consciousness that are still hotly debated.

AI as Your Creative Partner

So, does AI truly deliver creativity? Based on Boden's framework, I'd argue that AI demonstrates clear capabilities in combinatorial and exploratory creativity, with emerging potential in transformational creativity.

However, there are severe limitation in how creative activities are carried out by AI solutions today. I don't expect to see AI as replacing human creativity in a majority of situations, but rather as enhancing and expanding it. AI can serve as a powerful creative partner, generating ideas and connections that humans can then evaluate, refine, and contextualize.

In practice, the most exciting developments in AI creativity often come from collaboration between humans and machines. AI can help us overcome creative blocks, explore new possibilities, and challenge our assumptions. The biggest opportunity is in using AI to handle the heavy lifting of combinatorial exploration, freeing human creators to focus on higher-level conceptual work.

As we continue to develop more sophisticated AI systems, I expect we'll see even more impressive demonstrations of machine-based creativity that create pressure to deploy it in broader contexts. This is not without controversy. The rise of AI in creative fields amplifies concerns about job security if it is used to replace human creators in many areas. This trend also raises questions about ownership and originality of AI-created content, especially when AI learns from existing human works. Furthermore, some worry that AI might lead to less diverse and innovative cultural outputs due to its focus on replicating well known styles. These issues will inevitably heighten existing tensions between those advocating further AI technology deployment and others seeking to preserve human artistic expression in creative industries.

These are legitimate concerns. However, rather than fearing this development, the goal must be to seek responsible use of AI to expand the boundaries of human creativity and innovation. In the end, creativity – whether human or artificial – is about pushing boundaries, making new connections, and seeing the world in fresh ways. The promise and power of AI will be an important component in this endeavour.