AI’s role in Morgan, and numerous other creative endeavors, shows how far AI has come. Using techniques such as deep learning has enabled tremendous progress, but AI remains relegated to an assistant role—for now.
“What's interesting is that, compared to a lot of other machine learning techniques, deep learning technology is what's called a ‘generative model,’ meaning that it learns how to mimic the data it's been trained on,” explains Jason Toy, CEO of Somatic, a start-up focused on developing deep learning applications. “If you feed it thousands of paintings and pictures, all of a sudden you have this mathematical system where you can tweak the parameters or the vectors and get brand new creative things similar to what it was trained on.”
But even highly touted AI techniques have their limitations. “Creativity and where we started exploring with the Morgan movie is fascinating because deep learning isn’t the answer to creativity,” says IBM’s Smith. “We still have to define what creativity means. We know some of the attributes have to do with finding something novel, unexpected, and yet useful.”
“It’s easy for AI to come up with something novel just randomly. But it’s very hard to come up with something that is novel and unexpected and useful.”
— John Smith, Manager of Multimedia and Vision at IBM Research
By specifying teaching parameters for creativity, artists have gone as far as using AI to design sculptures and create paintings that mimic great works of art. For example, using the style transfer technique, artists can “teach” AI algorithms by showing them pictures of a style of painting like Impressionism to transpose photos and video to the same style.
These capabilities aren’t just relevant to fine art. “I see the whole creative industry from film to advertising and marketing using these tools to test out new ideas and accelerate prototypes,” says Toy.
Can AI learn to be creative?
Experts contend that we’ve barely scratched the surface of what is possible. While advancements in AI mean that computers can be coached on some parameters of creativity, experts question the extent to which AI can develop its own sense of creativity. Can AI be taught how to create without guidance? Can it truly understand what is beautiful, perhaps by looking at pixel arrangements or color palettes?
“Just a few years ago, who would have thought we’d be able to teach a computer what is or is not cancer?” says Arvind Krishna, Senior Vice President of Hybrid Cloud and Director of IBM Research. “I think teaching AI what’s melodic or beautiful is a challenge of a different kind since it is more subjective, but likely can be achieved. You can give AI a bunch of training data that says, ‘I consider this beautiful. I don’t consider this beautiful.’ And even though the concept of beauty may differ among humans, I believe the computer will be able to find a good range. Now, if you ask it to create something beautiful from scratch, I think that’s certainly a more distant and challenging frontier.”
Can we take what humans think is beautiful and creative and try to put that into an algorithm? I don't think it's going to be possible for quite a while.
— Jason Toy, CEO, Somatic
Experts point out that teaching computers to be creative is inherently different from the way humans learn to create, although there’s still much we don’t yet know about our own creative methodology.
“Many examples of creativity involve learning and exploring in a hierarchical style. Neural and multilayer network systems can help us construct different frameworks to better understand those hierarchies, but there’s much more to learn and discover,” explains University of Sussex cognitive scientist Margaret Boden, who also serves as an advisor at Stephen Hawking’s Leverhulme Centre for the Future of Intelligence.
“If you have a computer that comes up with random combinations of musical notes, a human being who has sufficient insight and time could well pick up an idea or two. A gifted artist, on the other hand, might hear the same random compilation and come away with a completely novel idea, one that sparks a totally new form of composition,” says Boden. She estimates that 95% of what professional artists and scientists do is exploratory and perhaps the other 5% is truly transformational creativity. A lot of the processes behind creative thinking is still unknown and Boden believes AI has a big role to play here.