Home / IT & Computer Science / AI & Robotics / Generative AI in Higher Education / Can generative AI be creative?
Can generative AI be creative?
In this article, Dr Martin Compton explores the definition of creativity to try to understand whether we should consider GenAI tools to be creative.
© King’s College London
In the previous step, you were asked how intelligent you think AI currently is and how intelligent it is likely to become. In this step, you will explore some philosophical questions about the nature of artificial intelligence, and about the creativity of generative AI.
Is AI creative?
If a text to image generator can create any image you can describe, where does the creativity sit?
The image below, for example, was created by my daughter using Midjourney, a generative AI tool that generates images from text prompts. Her prompt was only three words, two of which I had to look up: ’cyberpunk duck chibi’.
A Cyberpunk Chibi Duck, Inaya Compton, August 2023 via MidjourneyI used a reverse image search to identify similar existing online images, but few results came close. My daughter is artistic and creative but could not have drawn this. Of course, no one else has drawn it and I would not have thought to put those three words together, but is she being creative? Has the AI been creative? How far does this necessitate new terminology or a new way of defining creativity?
Similarly, if I ask a text generator to write a Haiku about a very niche subject, can I claim any credit?
I used the same terms in a Bing Chat prompt: ’write a haiku about a cyberpunk duck’:
A cyberpunk duck,In goggles and a black suit,quacks in neon light.
To be honest, charged with the same task I’d have spent a lot longer than the few seconds it took to generate it. Can I still enjoy it with that knowledge? Does knowing it is not human-authored diminish its quality or even my ability to dislike it?
The democratisation of creativity?
If the locus of creativity and how we might define it is as much a semantic question as it is philosophical, the secondary questions are perhaps more controversial: are these tools democratising creativity or is it no more than re-hashing the creative labours of the writing and art on which these generative AI tools were trained? Beyond ongoing and complex copyright disputes, what about creativity itself? Will it be dulled or deadened as a consequence?
In the context of education, we are obliged to ponder the critical question: what does this mean for those studying the arts or working in creative disciplines?
AI image generators can envision new building designs or swiftly produce hundreds of logo ideas, gradually honed through thoughts and words rather than the artistic materials (or computer assisted design software) of the architect or designer. Is this actually a continuation of a trend started when designers shifted from analogue to digital, or something new?
I’m reminded of a question one of my students asked himself in an MA dissertation years ago: should art students be able to draw? It was not a simple question to answer then, and generative AI makes it no easier now.
And what does it mean for what we teach and the jobs we prepare for? How far is resisting a realistic course of action? Will some creative endeavours come to be looked at the way we might now perceive traditional crafts? Ultimately, the question comes down to this: what do we lose and what do we gain?
Are there any new ideas out there?
It may be that your own answer to any of these questions could be shaped by pre-existing perspectives. Do you believe that it is possible to genuinely create something new? Have new PhDs, for example, really found a gap in human knowledge? Is a new artistic form really that new, or do you tend to side with Mark Twain who famously wrote to Hellen Keller when she was accused of plagiarism:
“substantially all ideas are second-hand, consciously and unconsciously drawn from a million outside sources, and daily used by the garnerer with a pride and satisfaction born of the superstition that he originated them” [1]
Taking (in Twain’s words): “ideas… consciously and unconsciously drawn from a million outside sources” sounds quite a lot like generative AI to me. What do you think?
Few would argue that the familiar tools we use daily to create written and visual outputs are themselves creative. They say a pen is mightier than a sword, but no one believes the pen or the sword has intrinsic strength until wielded in than hands of a writer, artist or warrior.
Can I compare thee to a brain?
Generative AI seems to be pushing us (though not consciously – or is it?) to think differently.
Brains are still much more powerful than computers, though the narrow and focussed capabilities of some computers may convince us otherwise. This can lead to imprecise comparisons between brain and computer functions, which in turn feed frequent tendencies to anthropomorphise generative AI (to attribute human qualities and characteristics to it).
The famous Turing test measures a machine’s ability to generate natural language responses that would convince a human evaluator. Large language models do now have this capacity. But the frequent assumption that an ability to ‘imitate’ is equivalent to ‘intelligence’ can strengthen tendencies to attribute human-like cognitive skills, and even sentience, that aren’t actually there.
Educators have a responsibility to understand and discuss our tendency to anthropomorphise AI technologies because it is potentially damaging. If we assume intelligence (and especially intelligence greater than our own capacities), then we are much less likely to question the outputs. It reminds me of news stories of people driving off cliffs or down impassable roads because their satellite navigation system ‘told them to’.
These stories are often presented as a portent of doom, as a signifier of our collective stupidity which will ultimately result in our inability to do anything for ourselves:
“In the year 5555
Your arms are hanging limp at your sides
Your legs got nothing to do
Some machine’s doing that for you.”Lyric from In the year 2525, Rick Evans (1969).
Now that you have completed this step, you have examined some philosophical questions about creativity and anthropomorphism that we should all consider when using (or when choosing not to use) AI. In the next step, you will return to more practical content, with an analysis of some generative AI tools currently available (in the UK), and what they can do.
References
- Twain M. Letter to Helen Keller from Mark Twain. American Foundation for the Blind; 2023. [written 1903; cited 3 October 2023].
Additional resources
If you would like to explore this topic further, here are some additional resources.
Russell M and Black J. Inside Sam Altman’s world, where truth is stranger than fiction. Insider. 27 April 2023.
Kasirzadeh A. The Socio-ethical challenges of generative AI. The Royal Society of Edinburgh. 2022.
Join the conversation
Do you think AI is creative?
© King’s College London
Reach your personal and professional goals
Unlock access to hundreds of expert online courses and degrees from top universities and educators to gain accredited qualifications and professional CV-building certificates.
Join over 18 million learners to launch, switch or build upon your career, all at your own pace, across a wide range of topic areas.
