Artificial intelligence has plenty of critics. But Sen. Bernie Sanders (I-VT) is quickly establishing himself as something more ambitious. He appears to be building the foundation for an entire anti-AI agenda.
His latest proposal may be the most extraordinary yet.
Earlier this month, Sanders and Rep. Greg Casar (D-TX) announced the Ban Artificial Intelligence Act, legislation that would permanently prohibit the development and deployment of artificial superintelligence in the United States. The bill proposes a pause on advanced AI development until Congress creates a new federal regulatory agency that would oversee the emerging technology.
Data center moratoriums to AI bans
Sanders’ new plan is only the latest step in a remarkably aggressive campaign against America’s emerging AI industry.
Back in March, Sanders joined Rep. Alexandria Ocasio-Cortez (D-NY) in proposing a federal moratorium on the construction of AI data centers. Sanders has cited concerns about electricity prices, water and land use, job displacement, and the possibility that increasingly advanced AI systems could eventually escape human control.
In June, Sanders introduced legislation that would give the federal government what he described as a 50% public-ownership stake in America’s largest AI companies, establishing something similar to a sovereign wealth fund. The plan would essentially nationalize half of each AI company, giving the government half of the board seats.
The newest proposal is particularly striking because it moves well beyond the environmental concerns that have fueled much of the opposition to data centers. The Ban Artificial Intelligence Act targets advanced AI specifically because of concerns the technology could become too capable.
And to be fair, some of those concerns deserve to be taken seriously.
Recent experiments involving advanced AI systems have produced troubling examples of systems behaving in unexpected ways, circumventing restrictions, and taking actions their developers did not intend nor anticipate. Even the companies building these systems openly acknowledge that increasingly capable AI could pose serious cybersecurity, biological, and societal risks.
These are legitimate concerns that should be acknowledged and addressed by those in this field.
But Sanders leaps from “this technology carries risks” to “Washington should possess the authority to decide when technology progress has gone too far, destroy offending systems, dissolve offending companies, and imprison the people who built them.”
That is an enormous and concerning leap for any government official to make.
When does research become a crime?
Under the proposal, a new Cabinet-level federal agency would be empowered to “supervise the destruction of prohibited artificial superintelligence.” Companies that disobey the law could face what Sanders and Casar call a “corporate death penalty.” Individuals found to be breaking said law could face prison sentences of up to 20 years.
That raises a question Congress should answer before threatening AI researchers with decades behind bars. Under Sanders’ legislation, where exactly does legitimate frontier AI research end and criminal activity begin?
OpenAI CEO Sam Altman has been remarkably candid about where he hopes this technology is headed. In an interview with Bloomberg, Altman said, “We are here to build [artificial general intelligence] and superintelligence and all the things that come beyond that.”
Would Sanders’ legislation require Altman to disavow that ambition?
What happens when OpenAI, Google, Anthropic, xAI, or a university research lab develops a system that performs better than humans across an increasing number of domains? At what point does ambitious research become an illegal attempt to create superintelligence?
Those questions become especially important when the consequences include prison or a “corporate death penalty.”
The uncertainty alone could have enormous consequences for America’s position in the global AI race.
Researchers, executives, and investors would have to constantly wonder whether the next breakthrough in AI would constitute a breach of the ill-defined federal restriction. Lawyers and compliance officers would inevitably become part of decisions that today belong primarily to engineers and scientists. This uncertainty alone could cripple this entire technological endeavor.
And that may not be the worst of the outcomes.
Beijing’s likely response
In this environment of increased red tape and confusion in the U.S., China would have every incentive to keep moving.
Sanders’ proposal attempts to solve that problem by directing the U.S. to pursue international agreements preventing the development of superintelligence anywhere in the world.
What happens if Beijing declines to enter such an agreement?
The U.S. could impose severe restrictions on its own researchers, but it cannot guarantee that the Chinese Communist Party will do the same. Even an international agreement would require Washington to trust that China had accurately disclosed classified research and agreed with American regulators about what qualifies as prohibited superintelligence.
China already appears to be exploiting every advantage they have in the AI race.
Earlier this year, OpenAI disrupted what it described as an influence operation likely originating in China that used AI to create social media content criticizing U.S. data centers while posing as Americans.
More recently, X identified roughly 200 accounts linked to a suspected Chinese bot operation attempting to persuade the public to oppose data centers.
Of course, this is not to say all those who oppose data centers and AI are Chinese propagandists, but it does show that China is actively attempting to disrupt the development of this emerging technology in America.
American policymakers should think very carefully before doing Beijing’s bidding.
America cannot regulate its way out of the future
AI is already a massively powerful tool with potential to become even more powerful. Truly dangerous capabilities may require government restrictions. But the U.S. should be extraordinarily reluctant to give Washington, D.C., the power to decide that human technological advancement has reached its government-approved limit.
LEAVE YOUR AI PANIC NEXT TO YESTERDAY’S NUCLEAR POWER OPPOSITION
The U.S. is currently home to many of the world’s leading AI companies, researchers, innovators, and investors. That advantage is enormously valuable, and it is far from guaranteed to last.
We should work to address AI dangers as they manifest without becoming so afraid of the future that we throw it all away.
Donald Kendal (dkendal@heartland.org) is the director of the Emerging Issues Center at The Heartland Institute. Follow @EmergingIssuesX.
Continue reading...
[ H/T Washington Examiner ]
