The systemic problems with bias in artificial intelligence utilized in the healthcare business have been addressed by new, industry-developed technical standards published by the Consumer Technology Association (CTA), according to a news statement issued on September 21. The consensus-based guidelines assist risk managers and AI product developers in determining the forms, origins, and management techniques of bias for AI technologies.
The ANSI-accredited standard from CTA, known as CTA-2116, describes the various types of bias, the potential locations for those biases throughout the data lifecycle, and effective data management techniques to reduce bias. The standard supports efforts by risk managers and AI product developers to reduce and effectively manage bias within AI systems in light of the potential dangers.
"The Consumer Technology Association has released new, industry-developed technology standards," said Gary Shapiro, president and CEO of CTA.
CTA-2116, titled "Artificial Intelligence in Health Care: Practices for Identifying and Managing Bias," identifies the phases of the data lifecycle where biases appear and classifies the various sorts of prejudice that can infect AI systems. It recommends "Good Data Management Practices" to reduce prejudice and ensure that AI technologies are created and applied with the highest fairness and accuracy.
The CTA's standards program, which has over 1000 active participants, develops guidelines for innovation in almost every sector of the technology industry. In addition to CTA-2116, the association has published articles on definitions and characteristics of AI (CTA-2089-A) and The Use of Artificial Intelligence in Health Care: Managing, Characterizing, and Safeguarding Data (CTA-2107-A), making early contributions to AI standards that support innovation.
Furthermore, CTA has recently released a National AI Policy Framework that provides a roadmap for legislative action to enable the United States to lead responsibly and openly on AI. The framework is accompanied by research that demonstrates consumer acceptance of the need for AI regulation and their belief that the U.S. federal government should take the lead by passing legislation that strikes a balance between consumer protection and fostering innovation.
The new industry-developed technical standards from CTA aim to address the systemic problems of bias in healthcare AI and provide guidelines for risk managers and AI product developers to reduce and manage bias effectively. With the publication of these standards, CTA continues to contribute to the advancement of AI innovation in various sectors, including healthcare.
