The rapid advance of artificial intelligence (AI) begs a daunting question – will we ever achieve ‘human-like’ behavior in computational systems? Simon Fraser University's (SFU) professor Steve DiPaola and his research team are developing a solution called the AI Empathic Painter, using natural interaction methods to enable users to converse efficiently, while highlighting two major human qualities— empathy and creativity.
DiPaola’s team recently showcased its work at a major AI conference—NeurIPS 2019— in Vancouver, which drew an audience of over 10,000 AI experts and enthusiasts. Their system enables visitors to approach and converse with a 3D avatar chatbot, which creates an artistic portrait of the visitors inspired by their emotions and personality via the team’s Empathy-based Affective Portrait Painter.
To achieve this, the researchers have combined their research in empathy-based modeling for AI character agents with machine learning models from the team’s creativity artistic system.
With a host of gestural, motion and bio-sensor systems, the team’s AI systems are designed to give coherent, empathy-based conversational answers via speech, expression, and gesture.
“Using our special system, the AI avatar can, through conversation, evaluate the user words, facial expression and voice stress to make an empathetic evaluation—just as a human would be able to, about someone they are talking to,” says DiPaola, a professor in SFU's School of Interactive Arts and Technology (SIAT), whose team includes post-doctoral researchers Nilay Yalcin and Ph.D. student Nouf Abukhodair,
Then the researchers take it a step further—using the Empathy-based Affective Portrait Painter to paint a unique portrait of the user, based on the empathetic evaluation. DiPaola’s AI artwork has been showcased globally in such museums as New York’s Museum of Modern Art and the Whitney Museum of American Art.
“The growing success of dialogue systems research makes conversational agents a perfect candidate for becoming a standard in human-computer interaction,” explains Yalcin. “The ‘naturalness’ of communicative acts provides a comfortable ground for the users to interact with. There have been many advances in using multiple communication channels in dialogue systems, simulating humaneness in an artificial agent.”
DiPaola’s and Yalcin’s extensive research on empathy in AI is also addressing issues in a variety of industries, including e-health. In a collaborative project with Canada's AGE-WELL initiative, a helper AI conversational bot is being developed to assist the elderly in staying independent at home. Other applications are geared to the entertainment industry.
After premiering the AI Empathic Painter system at the Vancouver NeuroIPS conference, DiPaola plans to travel to Europe to showcase the research in Florence in May 2020.
Formerly from Stanford University, DiPaola lead SFU’s Interactive Visualization Lab (iVizLab). The lab's goal is to make computational systems bend more to the human experience by incorporating biological, cognitive and behavior knowledge models. The lab creates computational models of human ideals such as expression, emotion, behavior, and creativity typically for gaming, sciences, arts and health fields.
In an early published study, DiPaola found clues about how the artist Rembrandt pioneered a technique to make viewers linger longer at his paintings. More recently, he focused on ‘blurred face’ technology targeting news photo anonymity.
Website/photos: https://ivizlab.org/research/ai_empathetic_pianter/