Chai Shai Chaishai with me

Schumer describes how the Senate will regulate AI.

On Wednesday, Senate Majority Leader Chuck Schumer unveiled an expansive, open-ended plan to regulate AI. Schumer has described AI as an unprecedented issue for Congress, one that has lawmakers “starting from scratch.”

According to Schumer’s speech in Washington, at least nine panels will be formed to identify and analyze the most pressing issues that must be addressed by rules on artificial intelligence. These issues include worker protection, national security, copyright, and the prevention of “doomsday scenarios.” According to Schumer, the first sessions of the panels, which will include of experts from industry, academia, and civil society, will take place in September.

Schumer argued that the resulting US answer may leapfrog existing legislative recommendations from across the world and that the Senate will turn to committee chairs and other vocal members on AI policy to produce bills reflecting the panel talks.

Schumer mentioned the European Union’s proposed AI Act, which was passed by the European Parliament only last week, when asked about other recent initiatives. “If we are able to pull this off in a meaningful manner, I believe we will be able to influence the course of AI development worldwide,” he said.

The speech is Schumer’s most clear-cut statement to date on a problem that has been bothering Congress for months because of how widely used tools like ChatGPT are: how to catch up or get ahead on legislation for a technology that is already in the hands of millions of people and is changing quickly.

With the potential to revolutionize how we live, work, and communicate, ChatGPT’s success has sparked a rush in Silicon Valley to create and release a new generation of generative AI tools. However, there are concerns that these same technologies may make factual errors, disseminate false information, and reinforce existing biases.

In stark contrast to the rapid development of AI, Schumer has advocated for a more methodical approach, with an emphasis on familiarizing legislators with the fundamentals of the technology and the concerns it poses before attempting to legislate. Beginning last week, he and three other colleagues have been holding a series of briefings for senators behind closed doors on artificial intelligence that will continue through the summer.

On Wednesday, Schumer seemed to address concerns about his rapid speech tempo.

“Many of you have been demanding action from us for what seems like months now,” he continued. “It’s been heard. Your message has reached me clearly.”

But he said AI is a new problem without a precedent for Congress to follow.

It’s not like labor or healthcare or military, he noted, where Congress has a vast past from which to draw. Even the experts can’t agree on the right kinds of inquiries for government officials to make. We’re basically starting from scratch.

To Schumer, his approach involves creating “a foundation for AI policy” that will accomplish “years of work in a matter of months.”

Schumer has stated certain guiding principles for this approach back in April, and he has elaborated on them here. On the day he formally unveiled the framework, Wednesday, Schumer argued that any law concerning artificial intelligence should prioritize fostering innovation above addressing concerns about threats to national security or democratic government.

Schumer argued for “innovation first,” but with safeguards, accountability, [democratic] underpinnings, and a capacity to explain results.

As unchecked AI threatens to destabilize election systems and make it hard to fully analyze an AI’s claims, Schumer argues that the last two pillars of his theory may be among the most significant.

Schumer was reserved in his comments, making no overt demands for action. He did admit, though, that it’s possible for a consensus to form that advises against big government interference in the technology.

A one point, though, was crystal clear: “Companies must be compelled to design an interface through which end users can readily ascertain the rationale behind and the foundation for each given answer generated by the system.”

However, it could be quite some time before the Senate unveils a full plan. Schumer estimated that it would take more than a few weeks but less than a year to complete.

He then remarked, “Months would be the appropriate timeline.”

Show More

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button