Artificial intelligence has created a storm that has hit the American people with full force: pulling in multiple directions.
The extraordinary potential for increasing the capacity for productivity and innovation is dazzling. The stock market is being driven upward by the vision of what can be achieved.
At the same time, there is resistance to big energy-guzzling data centers. But more fundamentally, there is the prospect that AI could lead to economic and social instability.
Even some of the key founders of AI are giving dark warnings about the dangerous potential for AI to become smarter than humans and to possibly inflict vast harms on humanity – visions that Americans have been primed to consider by numerous science fiction movies.
Beneath
those debates lies a key question: How active a role should the government play in regulating the development of AI?
To find out how Americans answer that question, my colleagues and I at the University of Maryland’s Program for Public Consultation (PPC) surveyed them nationally, and in 11 states and 28 House districts that have competitive congressional races in November's midterm elections.
One might expect the responses to fall along partisan fault lines. But we were surprised to find that robust majorities – in most cases more than two-thirds – of both Republicans and Democrats agree on numerous proposals for the government to play a very active role in regulating the development of AI.
The proposals came from legislation already introduced in Congress, along with proposals advanced by international organizations, including the United Nations and the International Committee of the Red Cross.
What do Americans want from elected officials on AI?
Here are some of the proposals supported by large majorities of both Republicans and Democrats nationally to regulate how artificial intelligence is developed and used.
Bipartisan majorities favor creating a federal agency devoted to monitoring AI, anticipating emerging risks, recommending best practices and enforcing regulations.
Creating a federal agency to oversee AI:
- Democrats 85%
- Republicans 79%
They support requiring AI systems that make consequential decisions – such as evaluating job applicants, approving health insurance claims or reviewing loan applications – to pass government-designed tests before they are deployed. Americans also favor allowing government audits of AI systems already making these kinds of decisions.
Requiring AI systems making critical decisions to pass government testing:
- Democrats 82%
- Republicans 78%
Agreement is even stronger when Americans are asked about protecting the public from deception. Large bipartisan majorities supported requiring AI-generated images, audio and video distributed publicly – i.e., deepfakes – to be clearly labeled. And when it comes to political campaigns, they favor prohibiting political campaigns from using any deepfakes.
Requiring publicly distributed deepfakes to be labeled:
- Democrats 88%
- Republicans 89%
Prohibiting political campaigns from using AI-generated deepfakes:
- Democrats 83%
- Republicans 85%
Americans have strong feelings about regulating AI internationally
The consensus does not stop at America’s borders. Republicans and Democrats alike support the United States working with other nations to establish international rules governing the most powerful AI systems. In particular, they would like to see an international agreement to prohibit AI-powered weapons capable of selecting and firing on targets without direct human control.
Working with other nations to regulate large-scale AI systems:
- Democrats 82%
- Republicans 78%
Working with other nations to prohibit AI-powered weapons that can fire without direct human command:
- Democrats 82%
- Republicans 81%
Too often, our politics assumes that difficult policy questions inevitably divide the public into opposing partisan camps.
Our research over many years has repeatedly found that when Americans are presented with specific policy proposals rather than partisan slogans, they often arrive at remarkably similar conclusions. Artificial intelligence appears to be another example.
The challenge before Congress is not determining what Americans want. On this issue, the public has spoken with surprising clarity. The real challenge is whether our political institutions are prepared to respond.
Dr. Steven Kull is a political psychologist and founder of the Program for Public Consultation at the University of Maryland’s School of Public Policy. For more than three decades, he has conducted in-depth research on public opinion and democratic governance.
This article originally appeared on USA TODAY: Most Americans want AI regulation. Will Congress listen? | Opinion
