IBM Watson Vice President and Chief Technology Officer Rob High said the company is heavily investing in voice- and emotion-recognition technology, according to an interview with The Australian Financial Review.
IBM has achieved an 85 percent to 90 percent accuracy rate for its cognitive conversational systems and is able to detect 52 personality traits from 300 to 500 words, Mr. High told The Australian Financial Review. He added this level of accuracy is similar to the rate at which average human beings can detect emotion.
IBM has also started to train Watson to detect facial expressions, according to Mr. High. The goal is for Watson to understand conversation through a combination of analyzing words, vocalizations and facial expressions, similar to how human beings interact.
But, Mr. High does not think Watson will ever be 100 percent accurate. "Frankly, I think it might be unrealistic to expect we'd get more accurate," he told The Australian Financial Review. "At some point, humans aren't good at detecting this either."
For Mr. High, this technological advancement may one day turn today's voice assistance into personal advisers.
"The value from these systems is how it can augment and amplify human cognition," he told The Australian Financial Review. "To help us, not to do our thinking for us ... We are not looking for an artificial doctor, but for it to help us make better decisions in a way that causes us to be inspired about things we wouldn't have thought about ourselves, or to let us understand a problem better."
More articles on health IT:
ONC revises patient engagement playbook
Epic, Baxter integrate pharmacy workflow systems
Anthem may pay $115M class action settlement for 2015 data breach