Democrats Renew Push For AI Regulation In Connecticut Amid Economic, Ethical Concerns

This post was originally published on this site.

FILE PHOTO: Sen. James Maroney rolls out a bill on the final night of the legislative session Wednesday, May 8, 2024, at the state Capitol in Hartford. Credit: Doug Hardy / CTNewsJunkie

As artificial intelligence continues to reshape industries and daily life, Connecticut lawmakers are once again proposing legislation to establish guardrails for its use

In 2024, many expected Connecticut to become the first state to pass AI regulation due to the leadership of bill sponsor Sen. James Maroney, D-Milford. While his bill easily passed the Senate, it stalled in the House when Gov. Ned Lamont threatened a veto, arguing that regulation should be left up to Congress, the Federal Trade Commission, and other watchdog agencies. 

Maroney has amended Senate Bill 2, which aims to ensure consumer protection, promote economic opportunity, and prevent algorithmic discrimination, and it seeks to address concerns from the business community, including tech companies and hospitals, who feel the broad nature of the bill may stifle innovation. 

During a news conference advocating for the bill, Senate President Pro Tempore Martin Looney, D-New Haven, referenced an op-ed featured in the Hartford Courant by former Lt. Gov. Kevin Sullivan, entitled Tame AI Before It’s Too Late, Connecticut, which described AI as “an unprecedented threat to society,” warning that AI has “the capacity to self-generate and self-sustain itself without human control.”

Senate Majority Leader Bob Duff, D-Norwalk, expressed concerns about data scraping, despite regulations existing, noting that major tech companies like Google and Microsoft are extracting vast amounts of data from the internet. He compared the current AI landscape to the early internet era, where the failure to regulate led to issues such as data privacy violations and misinformation.

“This country made a huge mistake back in the ’90s when we did not regulate or put parameters on the internet and also on social media. We have made huge mistakes as a country by not doing more to protect our residents, especially our children, by letting the market settle itself, which means that it’s really the wild west,” Duff said.

Maroney cited specific cases of harm caused by automated, AI-driven decision-making, including an elderly man in California being denied senior housing due to an AI mix-up with a different individual’s littering violation who had the same name, and Hospital systems that disproportionately limited care for minorities by making decisions based on a patient’s previous spending, and a 2023 landmark settlement involving iTutorGroup, Inc for utilizing automated hiring systems that discriminated against older applicants, specifically women over 55.

“The harms that we’ve seen from automated decision-making are not new,” Maroney said.

At a General Law Committee public hearing, Daniel O’Keefe, commissioner of the Department of Economic and Community Development and a former tech investor who was nominated for the position by Lamont in 2023, expressed concerns regarding the bill’s regulatory provisions, arguing that legislation is premature.

“My concern is right now we’re attempting to regulate fear of the unknown. We appreciate the intent to protect consumers, but being early in regulating a rapidly evolving emerging category with significant economic potential presents its own risk,” O’Keefe said.

In response, Maroney offered that AI discrimination is already well-documented.

“We are regulating the known. We already regulate AI in insurance and in government use. Why was it not too early to regulate AI for state government use in 2023?” asked Maroney, referring to Public Act No. 23-16. 

O’Keefe’s focus on economic issues follows reports that Gov. Lamont expressed his intention to prioritize artificial intelligence at the annual Fairfield University Capital Breakfast on Feb.19, noting its rapid growth as a key sector to Connecticut’s economy. 

Maroney pointed out that bio-tech is one of the most highly regulated industries, yet Connecticut is a leader in that field, rejecting the argument that regulation stifles economic growth. 

O’Keefe acknowledged that SB2 had the right intent, but urged the committee to consider Gov. Lamont’s AI bill, SB 1249, which would create a venture fund for AI and quantum startups at Connecticut Innovations, expand the state’s Open Data Portal with AI-ready datasets, and establish a regulatory sandbox for AI in insurance, finance, and health services.

SB 2 places a heavy focus on educating the public and workforce development, particularly with the establishment of the Connecticut AI Online Academy, which hopes to be utilized to develop a state technology strategy. The academy was previously launched in partnership with Google with an initial goal of 1,000 participants in the first year which was quickly met when over 1,700 enrolled in the first week. 

Maroney emphasized the need for AI skill development, citing projections that advancing technology will eliminate 85 million jobs while creating 96 million new ones. However, he noted these new roles won’t be direct replacements, requiring new training programs and education initiatives to equip workers with essential skills.

Maroney cited A National Bureau of Economic Research study that revealed women, specifically women of color, would be disproportionately affected as one-third work in jobs at high risk of full automation, compared to one-fifth of men. He also pointed to a report by Mckinsey & Company that found generative AI could widen the racial wealth gap by $43 billion annually if left unregulated.

“These next three years are critical before we’re permanently leaving people behind,” he said. 

SB 2 also seeks to crack down on pornographic images and videos that are created and shared by AI, particularly for purposes of revenge. 

Maroney detailed how Connecticut’s approach compares to other states that have opted to create independent regulatory frameworks in the absence of federal oversight.

“Right now, Colorado has already passed its AI bill. Virginia has also passed an AI bill, though their approach differs from ours. Rhode Island has introduced the same bill, and New York and Massachusetts have their own versions in progress. Maryland is closer to Virginia’s model. He further explained, the big difference with our bill is that we focus not just on protections but also on promotion and empowerment, while other states have tended to focus primarily on protection.”

SB2 is expected to have difficulties advancing past the Senate for the second consecutive year.