On AI, and the Fluid Boundaries of Creativity
A Brief Comparative Exploration of AI and Human Artistic Production
IMAGE: Jason Allen’s A.I.-generated work “Théâtre D’opéra Spatial”. Controversial first price winner in the digital category at the Colorado State Fair in 2022. Credit Jason Allen.
The dialogues surrounding the capabilities of Artificial Intelligence (AI) in creative fields are often characterized by a fundamental misconception. Namely, the erroneous conflation of the verb "to create" with the inherent quality of "creativity". While at a cursory glance this linguistic confusion may seem trivial, upon closer examination, it unveils a profound misunderstanding of the very nature of creativity, both in the realms of human and machine.
Of course this fact alone does not mean that generative AI by definition cannot be creative. In fact, I think it is about time to reevaluate our fundamental ideas and definitions of creativity in an age of generative AI. But now I am getting ahead of myself.
Why It Matters
The mixup between ‘to create’ and ‘creativity’ is more than a theoretical problem regarding AI in education. When we talk about skills that are uniquely human, we often mention critical thinking, agency, and creativity. Yet, for some reason, there is a widespread misbelief that suggests implicitly that the inference ‘AI created something; hence, AI is creative’ holds true. It does not.
Beginning with a brief etymological exploration, "to create" denotes the act of bringing something into existence. In contrast, "creativity" alludes to agency and purpose, and to the attributes of said creation, as posited in the "Standard Definition" of creativity (Runco & Jaeger, 2012). Public and academic discourses frequently stumble into the semantic trap of mistaking AI's ability to produce (create) as evidence of its inherent creativity.
Yet, before casting judgments on the authenticity of AI's creative potential, perhaps we need to mull over our assumptions about human creativity. Robert Shore's seminal 2017 book ‘Beg, Steal and Borrow: Artists Against Originality’ provides a useful lens to reassess our beliefs. Shore meticulously traces the rich tapestry of artistic creation, revealing that artists have perpetually borrowed, reinterpreted, and built upon the works of their predecessors (Shore, 2017). This historical lineage of art underscores the notion that originality, in its absolute form, might be more mythic than actual. Robert MacFarlane’s 2007 book ‘Original Copy: Plagiarism and Originality in Nineteenth-Century Literature’ also problematises the notion of originality and additionally provides the useful distinction between creatio ex nihilo (creating something out of nothing - typically by means of divine inspiration) and inventio (creating works of arts by inspiration from other artists).
To be clear, creativity and originality are not the same thing. In fact, this too can be somewhat a semantic trap, but suffice it here to say that creativity is typically associated with the broader generation of new and imaginative ideas, while originality is centered around how new or fresh an idea or work is compared to what already exists in a field.
In light of Shore's argument for the mosaic nature of human artistic production, perhaps it becomes imperative to re-evaluate critiques against AI's creative capacities. If human artists, celebrated for their creativity, have historically drawn from a communal reservoir of cultural motifs, can we then disparage AI for its analogous processes? Both entities, whether organic or algorithmic, engage in a continuum of repurposing and reinventing.
On a Slightly Philosophical Note
Herein lies the crux of the contention: while AI undoubtedly has the capacity to create, does it possess the discernment to produce works that can be deemed creative in terms of novelty and value? Concurrently, if human creations are an intricate web of borrowed influences and collaborations, where does one demarcate the boundary of genuine human creativity?
To elucidate this, it's paramount to understand the foundational mechanics of AI's creative processes. Most AI-driven creations are a result of neural network architectures, like GANs (Generative Adversarial Networks), trained on vast datasets. These networks generate outputs by identifying patterns and synthesizing new combinations (Goodfellow et al., 2014). Arguably, this mimics the human process of internalizing influences and subsequently synthesizing them into new expressions.
Yet, the distinction might reside in intentionality. Humans possess conscious intent, cultural awareness, and emotional experiences that influence their creations. These intangible qualities lend a depth to human creativity, rendering it rich and multidimensional. AI, on the other hand, lacks these experiential nuances. Its creations, albeit intricate and often beautiful, emerge from pattern recognition and data manipulation, devoid of conscious intent.
However, does this lack of intentionality diminish the value or novelty of AI's outputs? Or does it simply represent a different, albeit legitimate, modality of creativity? These questions challenge the very paradigms of how we perceive, evaluate, and define creativity.
Concluding Remarks
As technological dominance continues its crusade, the intersections of AI and human creativity will increasingly become focal points of academic and societal dialogues. While AI's prowess to create is undeniable, its alignment with our conventional notions of creativity remains debated.
Yet, as Shore's work suggests, perhaps our understanding of creativity itself is due for a recalibration. I for one would be very, very careful to say that generative AI can never be original or creative, even if - or rather especially because of - our current approaches to measuring creativity, as these may need to be revised for future generations of machine/human interaction.
In a world where the boundaries between originality and derivation blur, it becomes still more urgent to approach discussions of creativity from fresh and more nuanced perspectives than before.
Be well, and stay creative.
Whatever that means.
Literature
Goodfellow, Ian et al (2014). Generative Adversarial Networks. Advances in Neural Information Processing Systems 3(11).
MacFarlane, Robert (2007). Original Copy: Plagiarism and Originality in Nineteenth-Century Literature. Oxford University Press.
Runco, Mark and Garrett J Jaeger (2012). The Standard Definition of Creativity. Creativity Research Journal 24(1):92-96.
Shore, Robert (2017). Beg, Steal and Borrow: Artists Against Originality. Elephant Books.