Building on its 125-year legacy of helping society safely adopt new technologies, Underwriters Laboratories (UL), a global safety science leader, has joined as a member in the Partnership on AI (PAI), a multi-stakeholder organization that brings together academics, researchers, tech companies, civil rights organizations and others from across the world to better understand the societal impacts of artificial intelligence (AI).
"There's alignment between UL's desire to influence the safe adoption of AI and PAI's view that there are safety-critical applications of AI that need to be explored," said David Wroth, director of data science at UL. "AI systems need to enhance safety, provide fair judgments, and respect privacy and security."
Since 1894, UL has played a leading role in defining safety and helping innovations come to market. UL will participate in the Working Groups of the Partnership which are organized into thematic pillars, each focused on a specific area of risk or opportunity for the development of AI. By collaborating on the deliverables of the Working Groups, UL will extend the reach of its mission to promote safe living and working environments for people everywhere.
The six pillars include Safety-Critical AI; Fair, Transparent and Accountable AI; AI, Labor, and The Economy; Collaborations Between People and AI Systems; Social and Societal Influences of AI; and AI and Social Good.
UL has a particular interest in the Safety-Critical Working Group which explores the potential of AI to improve outcomes, enhance quality, and reduce costs in areas such as healthcare and transportation. The group is also working to help ensure that AI is implemented in ways that do not cause unintentional consequences where safety is put at risk.
"We are excited to work with the global science leaders at UL to help us in our mission to create AI for the benefit of people and society," said Julia Rhodes Davis, director of partnerships at PAI. "We have begun early work with UL and value their expertise in the Partnership's conversations on safety-critical applications for AI."
Some of the most noticeable applications of AI today include recommendation engines, natural language processing, pattern recognition (medical diagnostics) and autonomous vehicles. Less visible applications of AI include expert, interactive systems (customer service applications), intelligent learning systems, and predictive analytics.
Keeping pace with innovation, UL has begun to implement AI to serve customers, improve operations, and enhance its ability to research and examine issues related to safety, security, and sustainability. The standards development team is working with collaborators to draft UL 4600, a standard that will address safety principles and processes for the evaluation of autonomous products which will undergo additional consensus building before final publication.
"We bring a breadth of applicable expertise in adjacent science and technology areas that can inform the dialogue about AI," Wroth said. "We bring to the table expertise in machine learning and AI, but also IoT, sensor technologies, cybersecurity, standards, and education. We look forward to sharing lessons learned that can be applied to this new field of artificial intelligence."