Sen. Schumer’s 'Ambitious' New Approach to AI Regulation

U.S. Senate Majority Leader Chuck Schumer is proposing a new way to regulate AI since it is “unlike anything Congress has dealt with before.”

In a speech at the Center for Strategic and International Studies, the New York senator said he will not take the typical approach of holding congressional hearings that feature opening statements and 5-minute spurts of Q&A from each legislator to query on different issues.

“By the time we act, AI will have evolved into something new,” Schumer said. “This will not do. A new approach is required.”

While AI holds much promise in transforming life for the better, there are “real dangers” too of lost jobs, new weaponry, misinformation and others. As such, Congress has “no choice” but to acknowledge its impact, he said. “We ignore them at our own peril,” Schumer said. “With AI we cannot be ostriches sticking our heads in the sand.”

Having published a proposed framework in April, Schumer added a two-part proposal to the plan.

First, he is proposing the creation of a congressional framework for action called the SAFE Innovation Framework for AI Policy to support innovation while also adding guardrails. Schumer argued that without guardrails, AI innovation will be stifled or halted because of its risks.

This framework would seek to answer the following questions:

Schumer said that “without making clear that certain practices should be out of bounds, we risk living in a total free-for-all, which nobody wants.”

Is AI SAFE?

The SAFE in the SAFE Innovation Framework stands for security, accountability, foundations protection and explainability.

To advance security, there must guardrails around AI advances such that nefarious groups cannot use them for “illicit and bad” pursuits. Security also refers to job security, particularly displacement of white-collar, knowledge workers by generative AI. Schumer said care must be taken to not further erode the fortunes of the middle class.

To boost accountability, AI technologies must be prevented from being used to do such things as tracking kids’ movements, exploiting people with addictions or financial problems, perpetuating racial biases in hiring, violating the IP rights of creators, and other harms.

Protecting the nation’s foundations is also crucial, and AI algorithms must be aligned with the values of liberty, civil rights and justice. There should be ways to prevent AI from being used to subvert elections to protect the democratic process.

“Chatbots can now be deployed at a massive scale to target millions of individual voters for political persuasion,” the senator said. “And once the damaging information is sent to 100 million homes, it’s hard to put the genie back in the bottle.”

Finally, explainability is the most important feature of all because it makes AI transparent. “You should be able to ask: Why did AI choose this answer over some other answer that also could have been a possibility?”

“Without explainability, we may not be able to move forward. If the user of an AI system cannot determine the source of the sentence, or the paragraph, or the idea, and can’t give some explanation of why it was chosen over other possibilities, then we may not be able to accomplish our other goals of accountability, security, foundation,” Schumer added. “Explainability is, thus, perhaps the greatest challenge we face on AI.”

But Schumer stopped short of requiring companies to reveal their proprietary AI algorithms, which would be “harmful” and stifle innovation. Instead, they should develop a system that can explain simply why the AI model gave a particular answer and where it came from.

A New Process

The second part would see Congress create a new process to develop policies under the AI framework.

“AI moves so quickly and changes at near exponential speed, and there's such little legislative history on this issue, so a new process is called for,” he explained. “The traditional approach of committee hearings plays an essential role but won't on their own suffice.”

Instead, he proposed bringing together top AI developers, executives, scientists, advocates, community leaders, workers and national security experts to Congress for a series of AI insight forums to discuss how to approach developing AI legislation. The meetings will start in September.

After convening “the best of the best” in one room, the group should plan to do “years of work in a matter of months,” the senator said. “Our job as legislators will be to listen to the experts and learn as much as we can so we can translate these ideas into legislative action.”

Each forum will focus on a major issue in AI such as AI innovation, copyright and IP, use cases and risk management, workforce, national security, guarding against dooms-day scenarios, AI’s role in the social world, transparency, explainability and alignment, and privacy and liability.

“These insight forums are the first of their kind,” Schumer said. “They have to be the first of their kind, because AI moves so quickly, will change our world so dramatically, is deeper in its complexity, and lacks the legislative history in Congress that other issues have.”

Schumer said he will be joined by Senators Martin Heinrich (D-N.M.), Todd Young (R-IN) and Mike Rounds (R-SD) to lead this initiative. After hearing from the forums, the chairs of the appropriate committees are urged to develop related proposals: Senators Maria Cantwell (D-WA), Gary Peters (D-MI), Amy Klobuchar (D-MN), John Warner (D-VA), Dick Durbin (D-IL) and their ranking Republican members. Each chair and ranking member are asked to identify and explore areas to tackle.

Schumer also asked other senators to join his group who have spoken out about AI: Michael Bennet (D-CO), John Thune (R-SD), Richard Blumenthal (D-CT), Marsha Blackburn (R-TN), Josh Hawley (R-MO) and others.

All in all, Schumer seems to be singlehandedly trying to organize Congress around AI regulation. He himself recognized the scope of what he’s trying to do.

“Now, no question about it, this is all exceeding ambitious,” he admitted. “We’re going to work very hard to come up with comprehensive legislation. Because this is so important, we’re going to do everything we can to succeed. But success is not guaranteed. AI is unlike anything we’ve dealt with before."

This article first appeared on IoT World Today's sister site, AI Business.