Page 64 - Are You Future Ready?

Sideline

/ Prejudiced assistant? /

Data diversity in artificial intelligence analysis has
aroused a lot of attention.

For example, a research on cancer deaths over the
past 20 years in the U.S. drew an unreasonable
conclusion that black Americans had significantly
higher cancer mortality. It is due to its data samples
being highly concentrated on white males.

A startup in Toronto applied Al to a hearing
assessment for neurological diseases, such as
Parkinson's disease, they found that the technology
062
is only applicable to people who speak native
English and have a Canadian accent. Others may be
misdiagnosed as having a symptom due to a slow
response or unclear answer.

Behind the biased data, there is a biased researcher
and his biased value. A startup which developed
an AI medical device to diagnose eye disease
approved by the U.S. Food and Drug Administration
(FDA) for the first time, promised to pay close
attention to the data diversity. The widespread
concerns of academia, political parties, businesses
and the community is expected to bring order out of
chaos.