AI and the Arts - Media manipulators
- Mridula Anand
December 4, 2019
For those who believe they cannot dance, here is a way to tick that off your bucket list. A recent paper by researchers at the University of Berkeley provides insight into how they could transfer moves from a dancer onto a person who could not dance. The result - a seriously impressive dance by an untrained amateur. While Artificial Intelligence (AI) has thus far been able to recreate voices and faces, dance was a new high. Although it involved impressive deep learning algorithms with a multi-step transfer process, the product mapped the bodily motion from a trained dancer to a novice very convincingly.
To most, this seems improbable and futuristic. Not really. AI has become more accessible and no longer needs years of expertise. Further AI aids through the 'learning by doing' approach and is realistic enough to surprise the skeptics. This has been shown time and again in photography with apps like Rosebud morphing your face to a model's body or the Deep Voice software which can mimic any voice with minimum training.
A machine is trained to create a piece of media that's indistinguishable from the original and this certainly raises ethical concerns. A similar effort by Deepfakes resulted in the misuse of technology for unethical uses and quickly created alarm and rightly so. In the Bharatanatyam dance world, we are yet to see these manipulations take hold. Realistically it is possible by harnessing the power of AI to produce choreographies of Bharatanatyam pieces that are aesthetically pleasing and artistically accurate.As early as 1964, there were computer-generated random dance sequences in a University of Pittsburgh study as well as the once popular sequencing tool ChoreoGraph which served as a cue sheet for dancers. This doesn't mean AI will quickly replace choreographers or their choreographies. Each classical item is individual and abstract, and AI is limited in its creativity quotient.
However, this leap of data capabilities doesn't come without its caveats. Kinesthetic data can create choreographies that, combined with other dance forms, can create new choreographies with no human contribution. Would this replace choreographers? Can signature choreographies now be attributed to a machine with none the wiser? Would this help choreographers with limited capability compete with those who choreograph original pieces from scratch with their intelligence and not an artificial one?
Currently AI is a tool to complement, not replace, choreographers and dancers. It would be interesting to see how it evolves to an Indian art context and its acceptance in the artistic community.
A performing artist and a teacher for over 15 years, Mridula Anand is trained in the traditional Tanjore bani of Bharatanatyam. A trained Molecular Biologist and a lawyer, she has also released an explanatory CD on the theory of Bharatanatyam. She conducts the RMT Samskruti Music and Dance Festival every year. Her school Natya Silambam has a presence in Pittsburgh, USA, as well as Hyderabad.
Post your comments
Please provide your name and email id when you use the Anonymous profile in the blog to post a comment. All appropriate comments posted with name & email id in the blog will also be featured in the site.