Some Future Day

[00:00:00] ​

[00:00:00] Matt, thank you for joining Some Future Day. It's exciting to have you. Matt Frazier is New York City's Chief Technology Officer and runs the Office of Technology and Innovation, which was created by Mayor Adams in January of 2022. Matt, why don't you take a second? I mean, this is a major consolidation of efforts.

[00:02:49] It gives you a very, broad purview as it relates to very important topics surrounding technology for New York City and New York City residents. Do you want to kind of explain, to our audience what OTI is exactly?

[00:03:01] Sure thing, and Mark, absolute pleasure to be here. So, when the Adams administration took, seat in 20, in 2022, we took a hard look at what technology looked like across the city. And the mayor being very innovative and very focused on transformation. He's very well aware that he couldn't do any of these things without having a strong technology background. One of the things that we found across New York City government is that technology didn't have a central voice, a central strategy, and across the over 80 agencies that make up the city, Each agency had a tech arm that was focused on setting their own strategies. So, in order to make transformational change across the city, and in order to make that change feel uniform, we had to take a common approach of how we dealt with some of our business challenges and how we dealt with issues. So, to do that, we created one entity to rule them all, and focused on bringing all of the technology, both investments and strategies together. To making, to make sure that anything that we do going forward was focused on delivering products that can meet what New Yorkers need and not just focus on an agency specific issue. So the Office of Tech and Innovation took place and we used the city's former tech authority, Department of Information Tech and Telecommunications. As the cornerstone and then we built around that. So we brought together, Information Tech and Telecommunications, Information Privacy, Cyber Command, 311, the Algorithms Management and Policy Office. And the New York City Mayor's Office of Information Privacy all into a single entity which now builds the Office of Tech and Innovation. So for the city, we're the tech authority, we are the cyber authority, we're the privacy authority, the city's franchise authority, and pretty much anything that's tech adjacent now belongs to one organization.

[00:04:53] that's remarkable. And you're touching on a lot of words that I think are critical ranging from, you know, cybersecurity to privacy and beyond. an exciting initiative that you launched just in, March of 2023, at least phase one is MyCity, which I believe at that point in time, it was really focused on individual benefits such as childcare services.

[00:05:16] And now you're poised to launch MyCity. Business, which, has a very interesting take for the business community, and leverages artificial intelligence. Can you share a little bit about how the city is going to, release MyCityBusiness with an AI component, please?

[00:05:36] sure thing. So you touched on one of the very, one of the important things that the mayor ran on as part of his platform, which was democratizing access to information and making New York City government, more simple and work for the people that need access to government services. when we come out of the pandemic, one of the biggest things that we were focused on is getting the New York City economy. we started, and in order to do that, it required folks to either reopen businesses or people would go back to work in person, all the commerce, all the restaurants, everything else that we had running in the city, people would take advantage of.

[00:06:12] And people are reluctant to leave their home or to go to work when they're in a position where they don't have adequate child care. So we had a bunch of child care subsidies that the city offered. But the problem was, for the average New Yorker, they didn't know how to it. So we built a portal, that focused on streamlining the access for child care programs that were administered by several different agencies into a single place. Now very similarly, our challenges from a business perspective are the same. You want to open up a business in New York City, you have to deal with several different agencies. For example, if you want to open up a restaurant, you might have to deal with small business services, the Department of Buildings, the Department of Health. and a number of the Department of Parks, depending on whether you have outdoor seating, there's a number of places that where you would have to touch. So what we said as the next, forefront is to simplify the business experience in New York. There's no reason why when you want to do business in New York, you can't come to a single place and get all the answers that you're looking for. So added the business portal into the MyCity ecosystem, but we knew that wasn't enough. So we wanted to use, technology to help supplement the existing website and add additional features that could help the business community. So we went around, we spoke to a bunch of businesses in New York and what we heard was, although you have information available on a portal, it's not always clear what information applies to my specific case or something that I'm looking for. So we used, for the first time, we built a chatbot that's focused on taking in questions in real time. And providing answers that are contextualized by what the person is actually looking for. So now in the My City business portal, if you jump on it and you say, Hey, I'm trying to figure out how to open a restaurant, it'll tell you the agencies that you have to interact with, but then it'll also give you the other steps that you need post interacting with the, with those city agencies. So it's, a very intuitive way to get access to both city services. But it also simplifies and brings down the barrier so that even if you don't know what you're looking for, it gives you a place where you can go and throw a question out and it can help you solve a problem that in some cases you may not even know that

[00:08:22] That's great. Matt, did the process in training the chatbot, the AI, did that take a long time? And how did, how did you go about that?

[00:08:33] you have. So one of the things that we're very conscious about is any model or large language model that you adopt, we want to make sure that when you get an answer from it, the answers are authoritative and they're true. To the best of our, to the best of our knowledge. So the way that we've done that is we've partnered with Microsoft and we're using their OpenAI interface. And we basically took a model, a closed language model, we took all of our contacts, all of our 301 data, all of our business data, all of our restaurant data, and all of our, and our, the small business data, we put it in a single repository and we trained a model based on that. And then we had a bunch of analysts go through and test the model by throwing questions out to test not just the accuracy of the information that was provided, but make sure that the context that was coming back was also appropriate. So for us, it didn't take, it took us from the beginning of the year till now, about three months between starting, between the point where we start to test the model and where we deploy the model, and so far, the feedback we've got was significantly positive.

[00:09:39] just for fun, did you consider at all creating a chatbot in the likeness of a metahuman?

[00:09:46] So, funny enough, we are,

[00:09:50] Adams?

[00:09:50] Yeah, it's funny enough, we're working on a couple of things that may touch just on that line, to humanize the experience. And in addition to that, from a voice perspective, we've got a couple of things that are coming down the pipe that will, touch on exactly what you said.

[00:10:08] That's pretty cool. It's interesting because naturally when topics surrounding AI are raised, people are concerned about bias and all. Something that we worked on as an agency was bringing up metahumans that could be custom built by the user, so it takes on the likeness of an individual that the user can relate to.

[00:10:30] And it might be interesting Visually or aesthetically to bring that to the forefront for New York City residents.

[00:10:38] Yeah, so to get to that point where we're capable of doing something like that at a broader scale, is something that I, in an ideal world, we would, but at this moment, we think about like a process, much like human development. Crawl, walk, run, jump. we are in the AI space. as we jump out, especially in the customer service space, we are, we're not like infants crawling, but we're upright and we're walking at this moment and we're trying to get more, products out there.

[00:11:09] And the further we get along is the closer that we get to something like that. when you look at how AI's been used in other spaces like cybersecurity. We've had AI in a significant way in cybersecurity, embedded in almost every product for the last decade. And in real time, you look at the cybersecurity products make, it's, determining what processes and what procedures should it allow, and what things should it block. And in New York City, for example, from a cyber perspective, we see about 80 billion cyber events weekly. And because of AI and automation, we're able to distill that down to less than 20 things that someone looks at that someone looks the communication space we're branching into, but in other spaces we have more comfort.

[00:11:55] So that number, 80 billion, suspicious cyber events each week in New York City is remarkable. I read that's on par with what the United States Army Cyber Command, as well as the Department of Defense, is experiencing. Is that accurate, Matt?

[00:12:14] That is correct. The only entity that we know, across the continental United States that sees as much cyber activity as we do is US Army Cyber Command.

[00:12:24] Yeah, I was really surprised to learn that, your team really helped me dig into that. And it uncovered this whole concept of non kinetic warfare and how China and Russia as well are attacking the United States. But it seems like a lot of that is focused actually right here in New York City. So, your role takes on a certain defense mechanism to a certain extent as it relates to cyber security.

[00:12:52] And, to take the concept a little further, what I've researched is that, whereas the United States might use artificial intelligence, Surrounding cyber attacks, it's typically against the opposing government, but because we're in a democracy, when we have attacks from, let's say, China, it's targeting the private sector, like individuals and businesses.

[00:13:15] so like, for example, they might steal technology, they might steal trade secrets, they might monitor. decision making to anticipate what the United States will do. So in your role, Matt, do you feel like you're taking on a much bigger responsibility, specifically as it relates to this non kinetic warfare that I'm describing, beyond New York City?

[00:13:39] Yeah, so for us, cyber security and defense on the cyberspace is a practice of community. And it isn't something that any one entity can do in and of itself. Now, New York City, across the global landscape, perhaps has a presence that's unparalleled. And when you look at the amount of cyber activity we see, It's no secret why that's here. our mission and our job when we look at it from a cyber perspective is more than just protecting New York City, but it's about how we protect the greater homeland and the greater state. So, we worked in the first part of the, first quarter of the administration to launch a joint security operations center, and the joint security operations center was focused on doing just that. Looking at Cybert beyond just the city lens and bringing together the city, the state, and other cities across the state. Plus the federal government so that we can look at cyber collectively as, you know, as, a mission that we accomplished together. So the JSOC, which is located in one of, which is located at OTI Cyber Command, right? We now have, presence from both New York State. New York City, both police forces from New York State Police, New York City Police. we have partnerships on the federal level, both from the FBI and other areas. So now you have one center where you can report something that's happening. And in real time, we can work on not just looking at the incident and helping. Recover whatever has been compromised, but then we can also look at it from an attribution perspective. Who's the actor that, that, may be causing this particular activity and how do we protect not just the person that's been impacted, but other organizations that may be subject to the similar type of vulnerability.

[00:15:25] So when I look at cyber as a whole, it's not just us, it's us in

[00:15:29] Yeah, I mean, it's really remarkable. It's much bigger than New York City. And I think that most New York City residents, including myself, frankly, didn't realize that your administration has such a big, role in, really preserving our lives, in many ways. So, you know, if China is to put malware into, our critical infrastructure, what areas?

[00:15:52] Can they actually impact here in New York City? Like, what should we be concerned about, Matt?

[00:15:57] And we also have compensating controls in place should something happen that we, help respond to those things as quickly as possible. instead of looking at something that's theoretical, we can look at something that's happened in the recent past. So if you're not, familiar, circa 2020, there was a compromise, 2020, 2021, with Solowinds. where Solowinds is a network manufacturing, a network, services tool set. and what we found, was that the network tool set of the program had been compromised internally. They had a couple of internal employees that were compromised, they injected malicious code, and that malicious code, was embedded in their product and distributed to anyone that consumed the product, anyone that purchased it. And by impacting just one company, the malicious actors were able to infiltrate almost every U. S. intelligence agent. All right.

[00:17:03] It's a good example of how bad or malicious code gaming into the right place can hit almost everywhere. So for us, the things that we are all always concerned about is critical infrastructure.

[00:17:15] We're concerned about the water, right? Our water supply. And we're concerned about foods, banks, financial institutions. But over 90 percent of the things that people consider critical. They're not operated by the government, which is why security has to be a practice, a community. Think about your healthcare institutions.

[00:17:32] Think about your grocery stores. Think about all the things that the government doesn't control. if we only protect the government's assets, then we're only in a position where we can only keep part of what's critical safe. We have to make it open and inclusive to make sure that we help those that those that also operate critical infrastructure to help protect themselves, which is why in the next coming, in the next coming months, we're working on expanding Cyber Command to include bringing in partnerships with, small businesses and other businesses that all operate critical

[00:18:04] That's really very interesting. So that, and it also just underscores my earlier point as it relates to China using artificial intelligence and malware to attack the private sector and individuals versus the government entity, right? They're going after these infrastructural elements that in a free marketplace are run.

[00:18:25] by the private sector. I read actually that the Biden administration is currently hunting, this is a New York, from a New York Times, a recent New York Times article that the Biden administration is hunting for malicious computer code that it believes China has hidden deep inside the networks controlling power grids.

[00:18:43] Communication systems, water supplies that feed military bases in the United States and around the world. intelligence and beyond. Like why, like to what end? Why would China want to use malware to attack our infrastructure? Matt?

[00:18:59] So it's, hard for me to say specifically why China, but I can talk in the abstract and the abstract, why anyone would look to attack specific infrastructure is if during any conflict, you want to disrupt quality of life and how do you disrupt quality of life? You break a couple of things, the things that make people comfortable, you break communications. So you start with disrupting the way that your emergency response operations work and so that you can't get to. The places that need you when they need you, then you work on disrupting other things like quality of life, like water, food supply, power, right? You make people very uncomfortable, also making them very uncomfortable.

[00:19:40] You also limit their ability to communicate and coordinate. So when you look at why some of these particular assets may be threatened, why they may be targeted, it's, all focused on disrupting your ability to respond to anything that may happen.

[00:19:55] So how will you recruit the private sector and in turn. Educate individuals about how important it is to have this New York City slash small business, private sector cooperation. It's really, it seems to me like it's really critical, but yet people don't understand, some of the nuances involved with, technology, privacy.

[00:20:19] And these, cyber threats.

[00:20:21] So, again, I'll give you a real life example of how we've done this in the physical security space and how it translates over to the cyber security space. In the months following 9 11, after it occurred, New York City, the New York City Police Department created a program called the Lower Manhattan Security Initiative.

[00:20:40] Now, LMSI was focused. On bringing in sensory assets. So what I, by sensory, I mean like cameras, radiation detectors, chemical sensors. So if you operate a large real estate in the city and you had cameras on your facility. Could you join a, a coalition of the willing to share that information so that you could get better response to crimes as they were occurring or conditions that may be impacting your specific facilities. Now at that program was so successful that we were able to get over 500 partners that joined that program. Since then on the cyberspace, we have a working group that represents Big tech, it represents communications and power utility and finance and a bunch of other verticals like health where we share information in real time about threats that are occurring across the nation. Now, we think that the part of the value proposition isn't just giving us information, it's what do we do with it. So, being able to provide tool sets that help these businesses protect themselves. I think that's the key thing that will attract people to come and join the program. Shared responsibility, shared mission will be part of it, but they also need value out of being part of, a part of it. So it's got to be more than just information in the event that you have a compromise or something along that lines, knowing that you're not standing by yourself and you have someone that will assist and help. help you recover anything that may have been lost I think that's going to be part of the value proposition just like we saw in the physical security side if you provide access to cameras then that means we can give you better response when it comes to things that may be incurring

[00:22:20] So when you talk about, like, access to information and data, I think that's very interesting also with, the recent, ban that Mayor Adams put on TikTok as it relates to governmental official's phones, and computers, declaring that there's a cyber cybersecurity risk, like, do individuals in the public at large understand how social media and personal networks feed into these broader national security or local security threats?

[00:22:52] So I I think the public has a general sense of awareness, but, these specific impacts, they may not understand fully what it means or what, it could, what could happen. So with tools like social media and tools that give you a lot of content very quickly, and especially cater content to the things that you like.

[00:23:12] You have the ability to provide this, to use those platforms to spread misinformation very quickly. But in addition to that, if you're not careful about what applications that you run on your devices, you could passively be giving information about other things that you do that you may not have been willing to consent to if you knew that it was actually collecting that sort of information. So I think the products that we consume, the technologies that we use, we just have to be conscious about the risks associated with using those. And be very careful about what permissions that we give to applications or services that are running on our devices or that monitor our activity with other people.

[00:23:50] What was the reaction like, within the New York City Governmental agencies, when you said, look, TikTok is, you know, not fair play here anymore. Don't put it on a government device. Was it a mixed response? Were people upset about it? Did they feel like it was too intrusive into their personal lives? How did you, find people reacted?

[00:24:11] So I think we had a large amount of support, but of course, there's, This is New York. Getting everyone to agree on one thing is a little bit of a difficult job. But, for those that had concerns, we weren't placing any prohibition on their personal use of TikTok. We just said on government devices and for government use, deprecated that. And it's... It may not be a permanent ban, it depends on how TikTok responds and there's a number of things that are going on right now that may bring it into a state that makes it more palatable. But what we rested on is the fact that, not just New York City, but the U.S> government had placed a ban and they were very articulate around why and what concerns that they saw. And at the end of the day, we were doing something consciously. That would help protect New Yorkers and also protect the homeland.

[00:25:06] Yeah, it's very interesting. I understand that TikTok is actually banned in mainland China. Did you, are you aware of that?

[00:25:14] Yeah, I mean, there's a lot of things that are banned in mainland China. I mean, it's no big surprise to me. Also, when you look at how China operates from a communications perspective, even from content that flows in, they're very conscious and careful about what content flows in, even off the internet. So a lot of stuff gets censored as it makes its way into the country and it makes its way out of the country. So it's no, no big surprise that TikTok is banned.

[00:25:40] With TikTok, though, it gives China effectively the ability to, access biometric indicators, right, through a person's phone, camera, effectively, right?

[00:25:52] So I think that's a very broad statement that, it's certainly possible, but we can't say at this moment, or I can't say at that moment definitively, that's what happens. But what we do depend on is our partnership at a federal level. And if the federal government has concerns and they relay those concerns, then as good stewards, we have to respect it and in

[00:26:16] Do you think that individuals, just private citizens, should, refrain from having the app on their phone?

[00:26:23] that case. Yeah, you see, America is a country that's built on choice, and I think that what we can do is supply people with information and let them choose how they respond. and in terms of the use of TikTok and other social media applications, I think everyone... Understanding, the concerns they make, their, respective decisions.

[00:26:43] So, you know, I see as I drive around New York City now or walk the streets, it seems like there are more cameras out. And, I know that we've been using facial recognition, tools throughout New York City for various reasons for, a while. is it fair to assume that's That facial recognition AI has been at play in Manhattan for, some time now.

[00:27:10] No. So facial recognition, I think that people have this misconception that every camera you see that it's getting your face, that it's not the reality. So facial recognition is employed only by a trained set of people, within, our police department. And it's a very, it's a very Small number of people that have access to that sort of tool set. facial rec in general, most of the cameras that you see across the city aren't placed. And a place in a position where it's ideal to do something like facial rec. Remember for facial rec, most cameras that you see in the street are from the top pointing down at you. Facial rec, you really need a good shot from your torso up to the top of your head, right?

[00:27:53] And then from there you can, the algorithms are, much more successful at detecting someone. So for most of the cameras that you see, in fact all the cameras that you see, there's no camera that's running in the city where real time facial recognition is

[00:28:10] So, as it relates to facial recognition, then, I'm sure you've heard of Clearview AI. It's used by the FBI. It's being used in Ukraine. I think it's used in Miami, actually, their police department in a very limited way.

[00:28:23] It sounds maybe even analogous to the way it's used here in New York City. how are they using facial recognition AI, here locally?.

[00:28:32] So, in terms of the use of artificial intelligence tool sets and the use of technologies like Clearview or anything along that line, there's a myriad of technologies that we use to do facial recognition. and those technologies are administered by a select group in the NYPD called facial recognition team.

[00:28:52] Now, each one of the members of that team are trained on each tool that team uses. And only that team has access to the technology or the tools to do that. Outside of that group, no one else within the NYPD has the capability or in the city has the capability to run facial recognition. So whether that be Purview or any other tool

[00:29:12] So Matt, with, artificial intelligence, entering the space now, what types of issues do you anticipate New York City, will have as it relates to the government using AI and the people, you know, as it relates to perhaps privacy or any other issues?

[00:29:31] Yeah, so for artificial intelligence, I think the biggest challenge that we have is, ensuring that the content that's being provided is accurate and also ensuring that the information that, it's consuming and it's being learned, that the behavior that's being learned is also, ChatGPT. OpenAI had a very good concept of having that moral compass, that engine that basically flagged any content that violated its rule set. And has a human analyst that looks at it to say whether it should learn from that behavior or should. Not learn from that behavior. As we saw with very early instances of artificial intelligence, I believe it was circa 2012, 2016, somewhere along that lines, Microsoft had launched a chatbot that opens Twitter, and within 12 hours of being live, it had taught it to be biased, right? I think for us, making sure that any way that we employ artificial intelligence, that we do it in a way that's responsible, that we can control and ensure that the information that's being learned and provided is accurate. And that we don't do it in a way where that anyone that uses those tool sets, it inadvertently gives up private information that's used to tune some of these algorithms. So I think that those are some of the areas that we look for in that space.

[00:30:51] Matt, before you go, do you want to just say a few words regarding the success of Big Apple Connect? I'd like to get you to talk to that just briefly,

[00:31:00] No word. So when we came into the administration in 2022, one of the things that we saw was that we were still in the heels of COVID. We were looking at a potential, another shutdown for the city. And one of the things that we noticed is that over 40 of the people that live in public housing didn't have access to broadband. In some cases, the numbers were more stark than that, but the average is about 40%. And you think about that in a time where we just came out of almost two years of lockdowns, right? It's like you have kids that were sent home. They didn't have access to resources so that they could compete fairly in school.

[00:31:36] They didn't have access to social outlets so they can communicate with their peers and friends. They were just home with nothing. So in the first year, we launched Big Apple Connect and Big Apple Connect was focused on closing the digital divide. Especially across the state that government controlled. So early in the year we provided free broadband in all of the family homeless shelters. Now we wanted to take it into the space where not only family homeless shelters, if you lived in a house and are in public housing, you should have access to broadband like you have access to heat and hot water. Because now we've gotten to the point in society where broadband is a necessity. So within a year of being in office, we provided over a hundred developments with access to broadband. And as we came through the rest of this year, we closed the loop and now we're in over 200 developments that have access to broadband.

[00:32:26] So we've effectively provided over 300, 000 people with free access to

[00:32:31] broadband. And it's more free access to broadband in one year. Any administration has done across an eight year term.

[00:32:39] Congratulations. That's really quite the accomplishment. And I acknowledge how important that is to level the playing field across our community, which will inevitably trigger innovation, entrepreneurship, and more value and wealth within these communities. So great job, Matt. Thank you so much for your time today.

[00:32:58] I appreciate it.

[00:32:59] Alright, mark, absolute pleasure and hopefully.

[00:33:02]