Committee Chair Lemar: "The appropriate time to regulate an industry is at its onset."
HARTFORD — “When is it too late?”
So asked state AFL-CIO President Ed Hawthorne Wednesday during a public hearing on the explosion of artificial intelligence (AI) systems out in the wild and the path to reigning them in in Connecticut.
There’s a lot of good that AI can do, he said, but not without a steady hand to guide it. Letting the technology proliferate unchecked poses risks –– mass firings, discriminatory hiring, and data harvesting, to name a few –– that the state’s “most vulnerable” just can’t afford.
Hawthorne offered that testimony at the State Capitol in Hartford while speaking up in support of Senate Bill (S.B.) No. 2: An Act Concerning Artificial Intelligence. That’s the wide-ranging proposal introduced by New Haven State Sen. and President Pro Tem Martin Looney alongside dozens of his Democratic colleagues. The bill represents one of the upper chamber’s top priorities this legislative session.
The bill was the subject of an hours-long hearing Wednesday hosted by the state legislature’s General Law Committee, which is chaired by New Haven State Rep. Roland Lemar.
Supporters of the bill, like Hawthorne, argued that there’s no time for state legislators to sit back and wait for a regulation-allergic Trump administration or Congress to act. For years, D.C. lawmakers have failed to pass comprehensive legislation on AI , the bill’s boosters argued, and so now it’s time for Hartford to step in.
“AI is here and it’s advancing at great speed,” said Hawthorne, “and we need to make sure consumers and workers don’t get railroaded by failing to enact reasonable regulations.”
Critics of the bill, like state economic development Commissioner Dan O’Keefe, countered that the proposal as written constitutes regulatory overreach. They said it would unduly limit AI’s transformative potential for process improvement and economic growth.
“Rather than being a state that welcomes innovation,” O’Keefe said about this bill, “we become the only state in the region that resists it.”
S.B. 2 itself outlines new disclosure rules for “high-risk” uses of AI, like image generation and “consequential decisions” affecting employment, education, lending and credit, housing, legal and government services, and healthcare. If passed, it would lead to the implementation of new enforcement protocols for breaches of those rules, to be carried out by the state Attorney General.
It also proposes the creation of a number of new government bodies to smooth workforce disruptions and provide training in AI systems, including a “Connecticut Technology Advisory Board” and a “Connecticut AI Academy.”
The bill follows a slew of AI laws targeting discrimination, displacement, and data use passed in four other states –– California, Utah, Colorado, and Illinois –– last year.
The proposal has sparked pushback from industry and the governor, who early in the last legislative session threatened to veto a similar bill, which died in the state House.
"Do Not Hide Behind The Guise Of Innovation"
Majority Leader Duff and Prez Looney on Wednesday.
Wednesday’s General Law Committee public hearing dove deep into the mechanics of the AI regulation bill.
Over the course of the morning and into the early afternoon, the committee heard or received written testimony from nearly 50 labor, technology, and business leaders on the proposal, giving the committee’s 22 members an opportunity to absorb comments in support and opposition before finalizing the text of the bill for referral to the assembly’s legal and fiscal research offices.
The hearing made one thing clear above all: Not everyone’s going to be pleased with this bill. The ACLU of Connecticut, for example, had worked with the committee years ago to draft another AI law –– Public Act 23 – 16 –– which created a working group to study the technology. At the S.B. No. 2 hearing, the group’s executive director David McGuire commended the committee for taking the next step in regulating AI. Still, he said, the current text provides “overly broad exemptions for certain companies and technologies, limiting oversight.”
McGuire pointed in particular to a “loophole” in the bill allowing companies to deny any appeal of an automated decision if it’s not in the “best interest of the consumer.” The bill does not specify what “best interest” means, aside from providing one example, relevant for AI systems in healthcare: “any instance in which any delay might pose a risk to the life or safety of the consumer.” The ambiguity, McGuire said, “could allow companies to deny appeals arbitrarily.”
Representatives from the state’s healthcare industry don’t see it that way. The bill, according to Yale New Haven Health RN Jacqueline Blake, would make many of the hospital system’s AI applications “impossible and prevent the use of innovation to improve healthcare.” AI systems that the hospital has deployed, or hopes to in the future, like “optimiz[ing] patient scheduling” to reduce “no shows,” medication management algorithms, “predictive technologies” to guard against the misuse of prescription drugs, and “patient data analysis,” could be taken offline if the bill goes forward as drafted, she said.
According to state Department of Economic and Community Development Commissioner Dan O’Keefe — who leads the agency, which, under the current version of the law, would be in charge of spearheading a “regulatory sandbox” on AI policy for state offices — the bill imperils the state’s capacity to innovate and compete with its neighbors.
O’Keefe appeared at Wednesday’s hearing remotely from India, where he and Gov. Lamont are touring with the state’s venture fund Connecticut Innovations (CI). He urged the committee to back a bill proposed by the governor, Senate Bill No. 1249, which would inject capital funding for AI projects through CI, release “AI-ready datasets” compiled by agencies, and create a similar “regulatory sandbox,” instead.
In response to this and other testimony pushing the committee to walk back the S.B. No. 2 AI rules and regs and adopt a lighter touch or “wait and see” approach, the AFL-CIO’s Hawthorne did not mince words. “It is essential that we do not hide behind the guise of innovation to allow corporate developers and profiteers to increase economic inequality, curtail our rights, and undermine our democracy,” he said.