Can AI automate human creativity?

In May, the InParadiso Gallery in Venice opened the first solo exhibition of the world’s first humanoid robot artist, Ai-Da Robot. Ai-Da, originally created by the Oxfordiansan “international team of highly skilled and broad employees” paints with a robotic arm, applying paint in the same way a computer can apply pixels.

Ai-Da’s exhibition explores the role of art in artificial intelligence – where it fits and whether robots can really be creative. It is “presented across five connected spaces, [and] will explore the interface between human experience and AI technology, from Alan Turing to the metaverse, and will build on Dante’s concepts of Purgatory and Hell to explore the future of humanity in a world where AI technology continues to invade everyday human life ”.

When we think of creativity, we tend to think of it as quintessentially human. The idea that we could automate something like creativity or imagination not only seems strange, but also undesirable – why would we want to leave beauty, humor and other aspects of the human experience to robots?

In 1998 Margaret Boden, an academic at the University of Sussex, explored these ideas in her paper Creativity and Artificial IntelligenceShe outlined three ways in which AI techniques can generate creativity: “By producing new combinations of well-known ideas; by exploring the potential of conceptual spaces; and by making transformations that enable the generation of previously impossible ideas.”

She rejected the idea that creativity was so uniquely human that it should not be automated: “Creativity is not a special ‘faculty’, nor a psychological trait limited to a small elite. Rather, it is a feature of human intelligence in general. It is based on everyday abilities such as associating ideas, remembering, perceiving, analogical thinking, the search for a structured problem space and reflective self-criticism.”

As Boden pointed out, although people may be born with specific talents – an ‘eye’ for color or the written word, for example – the ability to turn these talents into skills is a process of understanding, practicing and learning. And so can a computer.

John R. Smith is an IBM fellow and lead at IBM Research who has participated in a number of projects at the “crossroads of AI and creativity”. He tells spotlight about a particular project from several years ago that had a “big impact” that used AI to help create the first-ever “cognitive movie trailer” for the horror movie Morgan

Content from our partners

Small businesses can be the backbone of our national recovery

Railways must adapt to how we live now

“I learn something new on every journey"

In a blog post he spoke of the challenge: horror movies are subjective, and one person’s idea of ​​sheer terror might not make someone else break into a cold sweat. “There are patterns and types of emotions in horror movies that resonate differently with each viewer, and the intricacies and interrelationships of these are what an AI system would need to identify and understand to create a compelling movie trailer,” Smith wrote. “Our team was faced with the challenge of not only learning a system to understand ‘what is scary’, but also creating a trailer that would be perceived as ‘terrifying and exciting’ by a majority of viewers.”

The project demonstrated the value of AI in aiding the creative process, allowing the team to create a trailer that would have all the ingredients for a successful, terrifying preview of the film. But, as Smith puts it, “the creative step of ultimately producing something new, unexpected and useful from the suggestions is still up to human creativity”. In other words, the movie trailer still needed the team to check the process and put it all together.

Richard Norton is an artist and co-founder of two creative companies working with AI: little giant and NFT Peeps† He has been working with machine learning in commercial marketing since 2017, explaining that using AI in the production of his artistic endeavors has “created whole new ways of thinking about creativity and created things that I could never do with solely human techniques”.

In this way, although Norton uses AI to inspire, he explains that he mainly uses AI in producing actual images. He explains that this “gets great attention and good commercial numbers for customers and gives creative prices”.

AI gave him the ability to produce art he could never have made before, but he doesn’t think that minimized his role as an artist: “Devalue? Certainly not. In any case, it has helped to take your mind to all kinds of strange, wild and beautiful places.”

Smith agrees: “Creativity is still purely human work. That said, AI can enhance human creativity by automating a significant amount of background work that has the role of preparing for and even creating precursors for human creativity.”

Jamillah Knowles is an ex-tech journalist and artist with a degree in computer science. She has spent much of her career researching technology, especially AI, and is now pursuing a master’s degree in illustration where she seeks to change the way AI is portrayed in the media. She has also created artwork with AI programs, most notably embroidery.

Knowles is a big believer in using AI in art and is not concerned that AI will devalue the role of the artist. Instead, she focuses on the value of art to the public: “It’s figuring out where we find the joy and intrigue in art and I think that’s up to the individual.”

All the issues with using AI in creativity are connected to broader ethical questions, she says. Knowles refers to bias in data sets, where the data passed to computers about what we consider “good” art may disproportionately represent one’s personal biases about talent, success, and their socioeconomic determinants, but emphasizes the artist’s personal responsibility to to amplify voices. Knowles also points to socio-economic barriers to arts programs. “Fortunately, many of these tools (AI art programs) are free online. But it would be better if there was some support and incentive in education so that the kids who were doing art could find those tools [more easily]†

Ai-Da only makes art through automation, but ultimately the team of Oxfordians is behind it. They programmed it and projected their vision, creatively, strategically or mechanically, onto the robot. Can we ever see a program that produces art that is completely independent of any human artist?

Knowles isn’t so sure. “It is possible,” she explains, “but [Ai-Da is] don’t make [art] to make themselves happy or emotional. It comes to a successful result. It has been instructed to do something. So at every step you have some kind of human interaction to make that happen.”

[See also: Inside Europe’s fight for ethical AI]


Leave a Comment

Your email address will not be published.