A.I. Wrote a Housing Bill. Critics Say It’s Not Intelligent.
These days it seems like the reach of artificial intelligence knows no bounds, from generating images, prose and jokes to automating various work functions. But the technological advances can carry hidden costs.
Earlier this year, a lawyer found himself in hot water after he used Chat GPT to write his legal argument — resulting in citations of imaginary case law and nonexistent trial decisions. The safety of Tesla’s self-driving cars, which relies on A.I., is under federal scrutiny.
Now, for A.I.’s next star turn: A lawmaker in New York said he consulted an A.I. bot to come up with ideas for new legislation.
The lawmaker, Assemblyman Clyde Vanel, a Democrat who chairs the Assembly Subcommittee on Internet and New Technology, asked the software to identify a gap in New York law.
The result is a bill, entered into the legislative record by Mr. Vanel, that would amend the housing code to require landlords to provide tenants with a copy of their lease upon request up to twice a year.
As part of the bill, which was first reported by City and State, Mr. Vanel disclosed that it was researched and written by artificial intelligence.
“We can’t be Luddites,” Mr. Vanel, a Democrat who represents Eastern Queens, said in an interview. “We have to understand these technologies, and figure out how these technologies can enhance what we do.”
Mr. Vanel used Auto-GPT, a program that is a more advanced version of the Chat GPT software that has sparked a plagiarism pandemonium in American high schools.
The purpose of the bill, the A.I. explained, was to empower tenants and familiarize them with their responsibilities, while fostering “a more transparent landlord-tenant relationship.”
Mr. Vanel said he was impressed by the software’s insight, adding that he made some slight alterations after some human refining and fact-checking. Nonetheless, the bill’s practical impact is somewhat murky.
Tenants usually receive a copy of their leases when they sign them, though there are instances in which they might need to ask for a replacement — to prove the terms of their original lease agreement in litigation, or while applying for benefits, for example.
“It’s not something we care about all that much,” said Cea Weaver, a campaign coordinator with Housing Justice for All, one of the state’s most vociferous tenant advocacy groups.
Frank Ricci, of the Rent Stabilization Association, agreed: “I don’t know what the purpose of the bill is, to be perfectly honest,” he said, adding: “Maybe the real story here is that artificial intelligence in some areas just makes no sense whatsoever. Maybe it’s not so intelligent.”
James Fishman, a lawyer with decades of experience in landlord-tenant issues, said that such a measure would only be helpful limited circumstances, when such a document actually existed, and the landlord themselves had a copy. And because it says nothing about enforcement, Mr. Fishman noted, it would likely only benefit those whose landlords were willing to comply without being compelled to do so.
“It’s kind of like a homogenized view of New York landlord-tenant, without taking into account there’s so many nuances and gray areas,” he said.
To be sure, the lack of housing, especially affordable housing, is a crisis in New York that has caused many to leave the state. And though the issue was repeatedly proclaimed to top the list of priorities in the last legislative session, Democrats failed to reach a deal that would address the numerous and often conflicting goals of tenants, landlords, developers and policymakers.
Linda Rosenthal, a Democrat who chairs the Assembly Housing Committee, was unimpressed by the software’s contribution to the housing debate.
“A.I. failed to identify and to reflect back what the housing crisis is all about,” Ms. Rosenthal said. “It’s certainly not people clamoring to make their landlord, in the law, send them a copy of their lease.”
Mr. Vanel defended the software, saying that the bill was not meant to be “groundbreaking.” He said that seeking ideas from A.I. was no different from accepting those shared by constituents, or advocacy groups — a single step at the beginning of a very long, very human, legislative process.
But John Pelissero, a scholar of government ethics Santa Clara University, urged caution, saying that while it was interesting to imagine a world in which laws were made based on the priorities and morality of artificial intelligence, such a gambit was likely to have ethical and political costs.
“If the voters have to make another decision in two years about whether to keep that person in office,” he said, “they should be able to do so based upon a reasonable expectation that what they’re observing and seeing is the actual individual that they elected.”
He added: “It’s what representative democracy is based upon.”