Manhattan Borough President Mark Levine assembled a group of New York legislators, academics, advocates and tech policymakers — including a representative from Google — a week ago to discuss how people use artificial intelligence, and whether government regulation is keeping up with the explosive new technology.

Attendees described the closed-door meeting at 1 Centre Street as a brainstorm session to figure out what roles local leaders and stakeholders need to take as AI technology seeps into government operations, education, jobs and elections. The meeting aimed to identify the pros and cons of AI entering each of these arenas.

“What impressed me the most is that local policy makers are quite aware of these new technologies like generative AI and large language models,” said Dr. Jeanette Wing, executive Vice President for research and professor of Computer Science at Columbia University, who attended the meeting. “They’re also quite aware of the concerns that have been raised by the media already.”

While AI technology can refer to anything from visual deepfakes to smart home devices, Levine said the meeting was motivated by the meteoric rise of the ChatGPT language model and other forms of generative AI.

These computerized content makers are rapidly growing in popularity, conveniently producing text and photos that are hard to distinguish from human creation. Since launching in November, worldwide usage of ChatGPT has grown to more than 1.7 billion monthly visits — according to the internet analytics company SimilarWeb.

The U.S. makes up the largest share of visitors, and globally, ChatGPT already rivals search engines like Bing and Baidu, which have existed for years. The New York Times reported last month people are using AI like ChatGPT to plan gardens, design spaceships and appeal insurance denials.

“We started to realize in January that this was so big and so complicated — and that it was going to impact the city,” Levine said of the meeting’s planning stage.

Around this time, New York City schools banned ChatGPT’s use in classrooms due to its ability to complete written assignments. Political watchers are raising concerns about how generative AI could be used to spoof candidates or enhance lobbying campaigns in coming elections.

Entertainment writers, who went on strike last week, want AI banned from writing scripts and creating source material, among other regulations to protect writers and artists. Federal data shows that NYC ranks among the metropolitan areas with the highest employment of entertainment and arts occupations. Levine said his office is able to get the attention of local experts, industry leaders and government officials, making it ideal for an AI summit on these issues.

“The advent of AI, and its iterations through services like ChatGPT, are an iPhone moment,” said Andrew Rasiej, chairman of the New York Tech Alliance and CEO of the tech collaborative CivicHall. “Every new emerging technology creates a disruption.”

In the case of the iPhone, the disruption opened perpetual access to computers, blossoming digital life as people know it: Order lunch, stream Taylor Swift or learn about particle physics anytime you want. But smartphones also contributed to the overhaul or demise of the music sector, taxi services and a slew of other industries. People’s lives were disrupted.

The borough president’s meeting delved into how to address the potential good and the potential harms associated with generative AI — but on a local level, according to Rasiej and other attendees: Can NYC government services benefit from incorporating generative AI? Does the city have a place to regulate or is it better left to federal and state governments?

One example revolved around using ChatGPT to guide people through food assistance enrollment forms and help city officials to process applications — a system that’s currently cumbersome and filled with monthslong delays, due to staffing shortages. Similar tools could be developed for affordable housing backlogs, Levine said, though such ideas would require strategic planning to train city employees on how to incorporate generative AI into their jobs.

Attendees said the discussion also focused heavily on what ChatGPT means for education. Chegg, an education tech company that specializes in online tutoring, saw a 40% drop in its stock after it announced ChatGPT was cutting into its usage. If students can write 90% of their essays using generative AI, it could fundamentally shift how schools measure learning. Do teachers then respond by no longer giving out writing assignments?

“Professors may not necessarily have the ability to determine whether or not you actually have the educational skills necessary to navigate the 21st century,” Rasiej said. But ChatGPT could also be used like calculators or any other educational tool, whereby students are trained to use the tech to enhance their abilities, rather than replace them.

“Maybe we need to pivot to machine learning?” Levine said. “Is there even one course in machine learning currently being taught in the New York City public schools?”

Wing, from Columbia, said if government agencies opt to incorporate AI into decision-making, they need to be transparent and make sure the AI systems are robust and fair. The city is implementing a law this year to notify a municipal job candidate when an automated system is used in their application process, and officials will be required to conduct bias audits on the system each year.

Any regulations would also need to be designed with enough flexibility to not become immediately outdated as the technology evolves, Wing said.

Levine said he hoped the summit serves as a launching pad for guidance or official recommendations around local governance and use of AI.

“There's kind of a race between the capability of these systems and our understanding of how to keep them safe,” Levine said. “I want people to understand the extent to which these tools are already being used. This is not a hypothetical.”