AC Interview with Sabri Sansoy
How should we define creativity in Artificial Intelligence (AI) when we’re still struggling to define creativity for humans?
Dictionaries certainly aren’t helpful. Their definitions often are circular --- “the use of imagination or original idea to create something; inventiveness.” It’s correct but hard to apply. Personally, I like Henry Rollins’ workmanlike definition of creativity --- “starting with nothing and ending up with something. Interpreting something you saw or experienced and processing it so it comes out different than how it went in.”
This processing of something to come out different than how it went in lies at the heart of creative projects that include Artificial Intelligence. Some of these projects fly under the radar, like a science fiction movie screenplay written by a recurrent neural network (RNN). Or a bus stop billboard in London that used AI-powered facial tracking technology to look back at the people looking at it the longest and incorporating that experience into the next design iteration.
While purists debate how to define creativity for humans and algorithms, a clutch of creative technologists are declaring victory and moving on. One such creator is Sabri Sansoy, a roboticist, a scientist and an engineer who has found himself at the intersection of AI, art and advertising.
Sabri has a formal background in astrophysics and technology but has found most of his work happening at the intersection of creativity and technology. Recently, he worked with IBM and Marchesa on the "Cognitive Dress", a fashion industry project where a dress changed colors in real time based on the sentiment of audience tweets. These sentiments were determined by IBM Watson's Tone Analyzer. Another project involved the Chevy Bolt electric vehicle. Sabri's group trained an AI system to recognize brochure icons anywhere on the sales material. At the dealership, a buyer can snap a picture of a brochure at any angle and the system will recognize it to serve up a video corresponding to that page.
He's also building robotic beach combers and underwater remotely operating vehicles (ROVs) that use Deep Learning and image recognition to identify visible and hidden marine pollution such as cigarette butts and toxic metal contamination.
We spoke about those projects plus the tools and practices that bring AI into the creative developers toolkit alongside graphic and interactive design.