During a Dec. 5 event hosted by The Brookings Institution, experts explored the topic and what the technology could bring and how to best address it from a policy perspective.
There are, of course, many ethical questions surrounding the use of emerging technologies in and by government, and these systems could have serious consequences for vulnerable populations.
The Blueprint for an AI Bill of Rights was the Biden administration's first step in addressing this issue, and it offers a resource that provides principles and guidance around equitable use of AI systems.
“The principles outline the kind of world we should live in,” said Sorelle Friedler, OSTP’s assistant director for data and democracy, during her opening remarks. “Leaders across the U.S. federal government are already taking action protecting workers’ rights, making the financial system more accountable, and ensuring health-care algorithms are nondiscriminatory.”
This follows other federal advances in this space, such as the creation of a National Artificial Intelligence Advisory Committee earlier this year and prioritization of AI research as directed by an executive order under the Trump administration. And while the World Health Organization has released guidance for ethical use of AI in health care, this bill of rights is the first comprehensive guidance of its kind from the federal level.
Alex Engler, fellow in Governance Studies at The Brookings Institution during the panel, explained that this bill of rights is unique in large part because it has the weight of the federal government behind it. His belief is that because the resource is thorough and application-specific, it will help drive governance as well as individual use.
And while this document is comprehensive, experts agreed that there are still gaps that must be addressed.
“I think what’s clear is that this document represents mile one of a long marathon,” said Harlan Yu, executive director of Upturn, “And it’s really clear that the hard work is still in front of federal agencies and in front of all of us.”
This resource offers critical guidance for technologies that are already being deployed by government agencies; but it is not without blind spots. Experts pointed to the exemption for law enforcement agencies that was included within this bill of rights as one such example.
“I think this Blueprint for an AI Bill of Rights is a step forward, but it’s not the end all be all,” Yu said.
Friedler stated that the hope of releasing this resource is that by putting the weight of the White House behind this effort, it will help begin the process of creating a road map to help create policy around AI use moving forward.
The U.S. approach to AI policy differs from that of other nations in some ways. For example, the European Union has introduced the AI Act, a law that would assign risk levels to different AI applications to guide deployments.
While legislation obviously differs from a bill of rights, there is a lot of overlap in terms of the actual standards being set. However, Engler noted the difficulty of the AI Act’s comprehensive approach, arguing that there are advantages to the U.S. taking a more sectoral approach to offer specific guidance for specific use cases.
When it comes to U.S. laws on AI, pieces of related legislation have been introduced. Most notably, there is data privacy legislation on the table, but there are also other bills related to things like transparency and sector-specific harms, Yu explained.
The solution will not be one single piece of legislation to address ethical use of AI in the U.S., Engler said, but rather, it will be an ongoing adoption of many laws that complement each other and work together to address the different impacts of this technology.
