As artificial intelligence has become more powerful, more and more attention is given to how we can ensure it’s as fair as possible. A recent project by researchers at MIT CSAIL suggests that the key might be to focus on how the data that underpins AI today is collected.
“Computer scientists are often quick to say that the way to make these systems less biased is to simply design better algorithms,” the researchers say. “But algorithms are only as good as the data they’re using, and our research shows that you can often make a bigger difference with better data.”
The research was able to not only identify data that could cause potential problems but also to quantify the impact that factor could have on accuracy levels. They then used this to show how different ways of collecting data could reduce the various types of bias they had identified, whilst still maintaining high levels of predictive accuracy.
from DZone.com Feed http://bit.ly/2GPMJJ5
No comments:
Post a Comment