Trained AI models exhibit learned disability bias, researchers say

A growing number of organizations are using sentiment analysis tools from third-party artificial intelligence (AI) services to categorize large amounts of text into negative, neutral or positive sentences for social applications ranging from health care to policymaking. These tools, however, are driven by learned associations that often contain biases against persons with disabilities, according to researchers from the Penn State College of Information Sciences and Technology (IST).

ai models analysis analysis tools applications artificial artificial intelligence bias biases disability health health care intelligence machine learning & ai negative organizations policymaking positive researchers sentiment sentiment analysis services social text tools