The current explosion of interest in the rapid development of artificial intelligence and the data centers that feed it has led to an equally rapid explosion of legislation nationally and in New York State.

The legislators' concerns – like those of the general public – are twofold. One is the technology itself, primarily the enormous amount of energy it consumes. The other is the uses to which the technology can be put, consumer profiling for price-gouging, for example.

Underlying it all is a greater or lesser degree of anxiety over the Digital Doomsday scenario, in which the machines decide they've had enough with human imperfection and kill or enslave us all.

Locally, Assemblymember Sarahana Shrestha has jumped into the fray in a big way, sponsoring or co-sponsoring several bills and hosting a public forum last week with state Sen. Kristen Gonzalez at Midtown Kingston nightspot Tubby's.

Gonzalez, who represents parts of Brooklyn, Queens and Manhattan and chairs the Senate's Internet and Technology Committee, came by in the midst of her statewide “NY Fights AI Data Centers” tour in support of the Responsible Data Center Development Act (S.10642). Gonzalez is lead sponsor of the bill; Shrestha co-sponsors its version in the Assembly. The measure has passed both houses and awaits only the governor's signature.

“I don't know what [Hochul's] going to do,” Shrestha said in a phone interview. “But the wisest thing to do would be to decide as soon as possible. I don't know why she's waiting.”

As passed, the bill places a one-year moratorium on state permits for new data centers larger than 20 megawatts of power usage and requires an environmental impact report from the Department of Environmental Conservation on AI's statewide, “charting the current and projected impacts on water, electricity, land use, tax revenue and incentives, and pollution, and other adverse impacts of data center operations in the state.”

Additionally, “The bill directs the PSC to create a separate and distinct water and electric utility rate class for large data centers to require that these facilities pay their full share of costs to the grid and water infrastructure” and “establishes renewable energy and energy efficiency goals for data centers, as well as a community benefits program that will fund community infrastructure and residential home upgrades for host communities of such facilities.”

Those last points are particularly salient as “U.S. data centers consumed more than 4 percent of the country’s total electricity in 2023, and by 2030 that fraction could raise to 9 percent,” per researchers at MIT who noted that “A single hyperscale data center can consume as much electricity as 50,000 homes.”

In her remarks at Tubby's, Gonzalez depicted AI as a black-box technology with unpredictable behavior, the rapid growth of which has spawned a number of fell consequences.

“These large language models like ChatGPT and Claude … continue to train on us, and they're imperfect, and we don't quite know how they work, “ she said. “In the same way that, if you code a program you can find the bugs, you can't do that with generative AI because it's training. You don't know where it's pulling from.”

Pointing to states that have failed to restrain AI's growth, “We see water being polluted, energy bills going up,” Gonzalez said. “These big tech companies are just coming in and making the local government sign non-disclosure agreements, so they can't even talk as public officials about how they're negotiating with them … Some of that started happening here in New York. So we organized.”

In her turn at Tubby's mic, Shrestha depicted a technology in the hands of corporate power driven by profit and with an infrastructure at risk of bubbling into collapse.

The moratorium – an expanded version of the current moratorium on just the largest data centers – “buys us time,” Shrestha said. “What is happening is Microsoft, Google, Amazon, Meta, are working with AI companies that want to accelerate time. They are trying to condense the [development] timeline into the fastest possible, the biggest possible, and they are doing it with ambitions that nobody asked them to have. They came up with these ambitions themselves.”

“We're also kind of in an AI bubble,” Shrestha said. “They've set this agenda and they have realized that to keep funding their research, they have to also keep making money, right? Because they need expensive chips. They need all this data input. They need all this energy. So now commercialization has become a necessary component for them.”

And if and when the bubble bursts? “We don't have a plan for what happens when we permit data centers and they fail to stay in business.”

Not that Shrestha feels the technology is inherently bad. “The goal of companies like OpenAI and their competitors is to just go as big as possible, as fast as possible, as energy consuming as possible, as data consuming as possible, because they're all competing with each other to be the first ones to have a major [breakthrough],” she said in a phone interview. “But actually if we were doing this technology purely for the benefit of the public, that is not the way we have to do it. There are ways to do it with smaller models, smaller data sets, high quality data sets to train machines that would be able to do assistance with healthcare and so on. We can totally develop that kind of technology without going the route that these corporations are going, which is just all scale.”

But the assemblymember doesn't see that happening without public oversight. “Basically, we don't believe that would happen without regulation, without taking the corporation out of the seat that has the main control,” she said. “As long as these four or five corporations are the ones controlling the whole development we don't believe that the scenario we described, where some good comes out of it, is even possible.”

Shrestha expects a continuing flood of AI legislation when the state legislature convenes again in January. As sole sponsor, she last month introduced the Options for Search Listing Act, which requires search engine operators to enable users to turn off AI-generated search results.

In addition to the Responsible Data Center Development Act, Shrestha is co-sponsor of another Gonzalez measure, one that would regulate the use of AI to deter discrimination by algorithm. As the bill's text states, “A growing body of research shows that AI systems that are deployed without adequate testing, sufficient oversight and robust guardrails can harm consumers and deny historically disadvantaged groups the full measure of their civil rights and liberties, thereby further entrenching inequalities.” Notably, the measure includes a private right of action, that is, state residents will be able to sue Big Tech for violations.

Shrestha's out-of-session schedule has been a full one, the AI event coming toward the tail end of her healthcare town hall series. After that, however, she plans to host a similar series of events focused on AI, utilities and data center resource strain, dates TBD.