As part of its broader effort to support the development of trustworthy and responsible artificial intelligence (AI), the National Institute of Standards and Technology (NIST) is seeking feedback by its extended deadline on September 10 on its new publication, "A Proposal for Identifying and Managing Bias in Artificial Intelligence." NIST will use the public feedback to inform the agenda for several collaborative virtual events in the coming months.
According to NIST, as AI’s applications have grown, so has society's realization that its results can be "thrown off" by biases in the data it is fed, which includes data that captures the real world incompletely or inaccurately. "The approach the authors propose for managing bias involves a conscientious effort to identify and manage bias at different points in an AI system’s lifecycle, from initial conception to design to release," NIST reveals. The Institute notes that its goal is to involve stakeholders from many groups both within and outside of the technology sector, allowing for different perspectives that have not been heard otherwise.
“Managing the risk of bias in AI is a critical part of developing trustworthy AI systems, but the path to achieving this remains unclear,” said NIST’s Reva Schwartz, a report co-author. “We want to engage the community in developing voluntary, consensus-based standards for managing AI bias and reducing the risk of harmful outcomes that it can cause.”
NIST's effort follows its recent announcement requesting feedback on Artificial Intelligence and User Trust (NISTIR 8332), intended to examine how humans experience trust as they use or are affected by AI systems. That NIST document, envisioned to complement active work on AI system trustworthiness, is open for public comment until July 30, 2021.
ANSI Supports AI Efforts
Like NIST, ANSI continues efforts to support AI efforts and progress in AI standardization. ANSI's 2021 report, "Standardization Empowering AI-Enabled Systems in Healthcare," reflects feedback from a 2020 ANSI leadership survey and national workshop, and pinpoints foundational principles and potential next steps for ANSI to work with standards developing organizations, NIST, other government agencies, industry, and other affected stakeholders.
See related news:
A Matter of Trust: NIST Proposes Evaluation Model for User Trust in Artificial Intelligence
ANSI Releases Report: Standardization Empowering AI-Enabled Systems in Healthcare