AI is a mixed bag. For all its great promise, this new technology comes with serious pitfalls.
As a country, we will only be able to reap the benefits of AI if we can ensure that the American people have broad trust that AI systems are working for them, augmenting, rather than degrading, human potential. Therefore, the overarching question for policymakers should be: How do we wring the benefits out of AI, like better medical diagnosis and treatment, while neutralizing its risks, such as algorithmic discrimination?
Amid a flurry of recent hearings and announcements, Congress appears keen to pass something having to do with reining in artificial intelligence. Since 2018, Congress has actually racked up an impressive bipartisan record of AI legislation, laying a foundation upon which policymakers can build to answer the questions above and others.
The challenge is that while technological change is rapid, Congress moves slowly.
As policymakers gear up for years of debate about comprehensive AI legislation, we risk a situation where — in the long interim between today and any enactment of some public law — nothing is done to channel AI in ways that maximize its benefits and reduce its harms. But neither should Congress rush to pass legislation that might have unrecognized consequences for a technology that’s becoming vital to our national security and national competitiveness.
Fortunately, there is a middle ground. Congress can walk and chew gum at the same time — while the country waits for comprehensive AI legislation, policymakers can still respond to the AI moment in ways that serve the dual mission of wringing benefits while mitigating risks.
Here are four consensus-building proposals Congress can swiftly pass to do that:
First, Congress should make clear that all civil rights laws apply to decisions made by algorithms just as they apply to decisions made by people — discrimination, whether by man or machine, is illegal. In doing so, Congress should also invest more in the National Institute of Standards and Technology (NIST) to expand its work to solve the thorniest of AI problems. NIST needs more resources for AI testing, evaluation and verification, including for identifying bias and discrimination. Fortunately, we already have a proposal to do just that.
Second, Congress should work with the Biden administration to give them the resources and authorities they need to build the National AI Research Resource (NAIRR), a set of shared computing and data resources made available to researchers throughout the United States. You shouldn’t have to work in Silicon Valley to get access to the tools necessary for the next generation of AI innovation. By democratizing access to AI research tools and computing power with the NAIRR, we can help ensure more researchers are participating in efforts to make AI safer and more trustworthy.
Third, Congress should pass the Deepfake Task Force Act, which narrowly missed being included in last year’s national defense authorization despite enjoying unanimous support in the Homeland Security and Government Affairs Committee. Deepfakes are hyper-realistic AI-generated images or videos depicting events that did not occur, which are used in many concerning ways from pornography to elections. Involving collaboration between industry and government, this bill would help develop standards to ensure the provenance of digital content, thereby reducing the risks associated with deepfakes.
Fourth, Congress should build on the success of the National Security Commission on AI (NSCAI) by creating a new Commission on the Future of Work. The NSCAI was a successful nonpartisan commission created by Congress in 2018 to develop and recommend policies to hone the United States’ competitive edge concerning AI. Thanks to the NSCAI’s practicality and rigor, many proposals have become law, including their idea to build the NAIRR.
Undoubtedly, AI will transform our economy, including the nature of work. For that reason, Congress should convene experts with various viewpoints and backgrounds to develop concrete policies to ensure that our economy harnesses AI in ways that create jobs, protect the dignity of workers and promote overall national competitiveness.
For the past few years, and under presidents of two different parties, Congress has passed a number of bipartisan AI-related laws. There is no easy red versus blue partisan breakdown when it comes to something as new and transformative as AI. This is an American challenge. The more Congress can continue to work collaboratively and build consensus around quality AI policy proposals, the easier it will be to tackle the ever more complex challenges posed by this technology.
The evolution of AI and its applications will not wait for policymakers. Congress needs to do its part in developing sensible guidelines now.
Rob Portman, a fellow at the American Enterprise Institute and former United States senator from Ohio, cofounded and cochaired the AI Caucus in the Senate.
Sam Mulopulos is currently an Ian Axford Fellow in Public Policy and a former senior staffer to Portman.