With Artificial Intelligence (AI) going mainstream, it is forcing every organization to rethink strategy, hiring priorities, and IT operations among others. How are Government Stakeholders and IT Leaders getting their organizations ready for the impact of AI?

Contributors

• • Doug McCollough, Chief Information Officer, City of Dublin
  • Seth Wainer, Chief Information Officer, City of Newark

Download podcast

Apple Podcast, Google Podcast, Spotify, Pandora, iHeartRadio, SoundCloud, TuneIn, and Stitcher. Find other syndication channels here or search CIO Talk Network podcast on any other app.

Explore More

• • Steps to Intuitive and Integrated Government
  • Smarter Government through Big Data
  • How are City and County Governments Minimizing Risk to Build Trust?
  • How the Public Sector is Utilizing AI?
  • Steps to Intuitive and Integrated Government
  • Transforming Government Procurement and Contract Management
  • How is the Government Preparing to Defeat Global Cyber Crime?
  • Government and Technology
  • Government eDiscovery Needs and Challenges
  • Secure Government and International Borders

Transcript 

Sanjog Aul [00:01:01]:

Hello, and welcome to this segment on CTN. To learn more, please visit ciotalknetwork.com. And the topic is Getting Government Ready For AI  , which is artificial intelligence. And our guest for today’s show is Doug McCollough, Chief Information Officer with City of Dublin, and Seth Wainer, who is Chief Information Officer, City of Newark. So, Doug, how are you?

Doug McCollough [00:01:26]:

I’m doing great. And I should clarify that’s Dublin, Ohio, just in case someone thinks that I have moved to Ireland.

Sanjog Aul [00:01:32]:

So if we are not going to Ireland then, yes. Yes. Alright, and we have Seth as well. So, hey, Seth. How are you doing?

Seth Wainer [00:01:39]:

Great. Good morning.

Sanjog Aul [00:01:42]:

Good morning to you. So we are talking about artificial intelligence here. And when you try to talk about something which is so far out there and in context of government, it could become a very interesting conversation. But it is not to be ignored because at least in the private sector, artificial intelligence is showing promises. It is going mainstream. People are using it to understand how to serve and understand and serve the customer better. And frankly, for government, you’ve got constituents and you’ve got citizens. You’ve got to serve them better, but is there a place for AI? So let’s start, with the first question here. So I’ll start with you, Seth. If you look in the context of government, do you think AI has a place today the way we envision governments functioning today?

Seth Wainer [00:02:40]:

Yeah. Certainly. I think that artificial intelligence is, if you think of it not in terms of the very coolest stuff that science fiction has provided us, but rather just software that’s smart and maybe voice responsive, I think that that definitely has a place in government today. I’ll give you one easy example: 311 call centers are kind of the norm for cities that deal with lots of non-emergency problems or want to offload items from their 911 call centers. And I think that artificial intelligence has a really terrific role to play in that space in parsing what exactly the ticket is that a particular resident might have and getting it to the right department quickly. I think that would really help large cities, medium-sized cities, even small cities provide really relevant services to all the residents.

Sanjog Aul [00:03:30]:

So to that end, Doug, when you look at Dublin, Ohio and then let’s take your city for an example. You may have such 311 call centers and if you put AI, yes, you might be able to shave a few pennies or maybe a little more than pennies. But is this truly supposed to be an efficiency play?

Doug McCollough [00:03:51]:

Not necessarily. I will say that the City of Dublin does not have a 311 call center, but I have worked for cities that do. And I want to jump back on also what you mentioned about the way we think of government today. We should make the point that government is changing. And the organizations that you may envision in terms of a city hall and the bureaucracy, a lot of that is changing. And what used to be the bureaucracy, you can picture papers moving across the desk. A lot of that has changed just as our citizens are now doing much of their business on smartphones. They expect to walk into Target or Walmart or call an Uber. Those expectations have moved into local government as well. So my take on artificial intelligence is that we’re looking for our machines to begin to do things that we cannot do as humans. We cannot process that much data. We cannot move that fast. And that’s not something where we’re shaving pennies off. More so it’s an opportunity to take on new services that really are not feasible for us as humans. We need a machine that can be lightning fast. That could be security. It could be analysis of crime statistics. It could be traffic management. There are changes that we can make more automated that only a machine can do. And so I see this more as an expansion of services opportunity than just a cost control opportunity.

Sanjog Aul [00:05:23]:

So the way you explained, Doug, it looks like a back office, not exactly an efficiency play, but a rethinking play where the way that feeds the data to a police officer in the field or to a clerk in a town hall, you wanted to change that. And is that where it is going to be done, and would you think the citizens would notice? Or that’s not this is what I call a backstage hero. So artificial intelligence becomes a backstage hero. Is that where you envision it’s gonna go?

Doug McCollough [00:05:56]:

I think that’s one opportunity, but I also think there’s a front office opportunity. Our residents expect to interact with an organization with a lot more efficiency. And so there can be some front-office applications. We talked about 311. People may not notice that they are interacting with an AI. But I do think that some people may, particularly if we tell them. We’ve got to do a better job at describing to our citizens: this is what’s happening; this is the technology that you are benefiting from right now. And some of that may be front office and customer-facing.

Sanjog Aul [00:06:37]:

So, Seth, when you look at and based on Doug’s response, it looks like there are larger opportunities. So if I were to give you a white sheet of paper okay, Seth. Go ahead and draw the potential, new government which could be enabled using AI. It may take time. It may be phased. But what would that look like?

Seth Wainer [00:07:00]:

I would think that it would involve a lot of different endpoints and software modules with various APIs on top of them feeding into a central repository of sorts, and intelligent decision making coming out of that repository. So that’s

Sanjog Aul [00:07:22]:

At the technical level. What’s the — what will the world see? What will the experience be for the clerk who’s at the town hall or for a constituent?

Seth Wainer [00:07:32]:

I think ideally for residents, you’re going to see faster, more relevant services and a more responsive government. That’s the idea. People want responsiveness. If your garbage isn’t picked up, I don’t just want it; I want it picked up today. If there’s a problem, I’d like the problem to be addressed. If people are parking all over my sidewalk and there’s a problem, and you’re getting complaints about really difficult narrow streets because people are parking inappropriately on both sides, then the broader first step is you want to be responsive if you want to address the issue. Then artificial intelligence might help us determine those curbs need to be repainted. Those curbs haven’t been repainted in a while, and we need to paint the yellow lines a little more brightly so that folks are parking the proper distance from a stop sign. And if not, they’re getting a ticket. So I think the responsiveness of government, how our government responds to its residents to do things correctly and systematically, is ultimately what everyone looks for in government.

Sanjog Aul [00:08:36]:

So, Doug, Seth has a point where some of the areas that we will touch, but all the examples perhaps I got from both of you look like we are leaning towards using it to make whatever we have today a little more efficient and effective. Is there any disruption possibility? Because you’re using something which has not been used before, and the curves can be painted perhaps slower. The garbage is being picked up. My own garbage gets picked up every Friday or Saturday, and I haven’t seen something really changing in it. So it suddenly starts getting more accurately or exactly at 7 AM in the morning. I really care less. So why would I spend money? Why would I get everybody all riled up about this if there is not a major difference? So is there a disruptive aspect in the government which it could bring which is not happening today?

Doug McCollough [00:09:26]:

Sure. I think that disruption happens while we’re not looking. Oftentimes, we notice disruption after it’s already been done. I think that’s going to happen a lot in government. Our citizens expect us to be constantly getting more efficient without them necessarily having to be aware of each new efficiency. Some of the efficiencies that we are experiencing can pay for some of the improvements or the investments in AI. So I don’t think that citizens necessarily want to follow the numbers like that. But I do think we are changing the way we interact with our budgets and the way things are funded. I do want to add also to what was just said: a lot of these benefits are multiple. Just as you can use an AI to help you with planning — we talked about the curbs or do our parking processes need to be adapted — you can also use them for day-to-day operational things. A camera can cover an entire block of parkers or drivers in a way that a human cannot. When we begin using a camera to observe parking habits, we spend less on those individuals who are walking the street, taking down license numbers or using an app to do those things. Now a machine can do that. When we do that, we also can use that data in a completely different back-office way. So the benefits are multiple and not simple. That’s a disruption, but most people would not see it as a disruption because of the way we traditionally look at government.

Sanjog Aul [00:11:13]:

Would you, Seth, be able to point to a pain area which could be effectively solved by AI and is a chronic issue today?

Seth Wainer [00:11:25]:

Yeah. It’s difficult to make predictions, as Yogi Berra said, especially about the future. But I would say some cities deal with different types of problems than others. Especially, we’re talking about urban poverty, garbage collection, abandoned properties, land use, and how foreclosures affect cities like Newark or Detroit. I would hope that as we get really, really good at the data, we get really creative with land use and management, and we can address more rapidly and turn through properties from a position where they’re not being used to their optimal economic value to a position where they are. I don’t think there’s any shortage of residents who are always interested in investing, buying, and improving property. But land use, especially in cities, is far behind where the Ubers of the world are, where you can call the car over and you’re good to go. Land use is really tied up in a lot of red tape. So I would think that if you’re looking for a seismic shift, moving around and allowing for the better use of land in cities would be a really seismic shift.

Sanjog Aul [00:12:44]:

Based on what you just said, Seth, maybe the next step is to talk about the actual implications of AI besides, of course, you adding value. Doug and Seth, both of you shared some input. So let’s take a quick break, listeners. Let’s come back and let’s talk about the implications, which is social, legal, and ethical implications of AI in the context of how government serves its citizens and keeps them safe. And then, what is the dark side, if at all, that you see today if we start overusing it? And let’s define the word overuse of AI. Please stay tuned, listeners. We’ll be right back.

Sanjog Aul [00:15:32]:

Welcome back. So, the discussion here that was going on is related to what are the possibilities, how disruptive can AI be for the government. That’s great. So now suppose we are saying, yes. We want to embrace it and there is value. We have to before we charge ahead or whatever we are doing today, we have to look at the social, legal, and ethical implications. So, Doug, I’ll start with you, and let’s inventory the challenges which are very obvious in these three areas, social, legal, and ethical.

Doug McCollough [00:16:08]:

Sure. I think this scares people. It just makes people nervous. I think people have a mental image in their mind of a computer that’s humanized as opposed to an algorithm. And we’re well beyond an algorithm here. But people imagine their local government, I’ll just speak on local government, as really touching their lives. It’s close to them. They share so much with their local government. Part of the ethical aspects here is we are the stewards of a significant amount of privacy. We touch people’s lives very carefully. Seth mentioned land use and we have a lot of data about people’s homes and what they’re paying for things. We may have a lot of health data. We have a significant amount of data. The value to the community is using that data, so they recognize that. But they also question how their data can remain private. I think there are also aspects of people hold governments much more accountable than they do private corporations. They tend to forgive, in my opinion, corporate activities in the interest of the profits of that corporation because they know they’re getting the service. I don’t think that we have done that yet with AI in a government setting and selling the public on the concept that the benefits to them are tangible and significant. 

Sanjog Aul [00:17:43]:

You say a very interesting word, selling. And so, Seth, coming to you, people love to buy, but they hate to be sold. So instead of us selling, what would people want to buy as citizens? Have we polled them? What are the insights that we’ve garnered from there?

Seth Wainer [00:18:01]:

Unfortunately, it’s a really good question. I don’t have a terrific amount of data-driven insight into what residents are fundamentally looking for. I think the easiest way to see what residents are looking for in some ways is probably to look at the policies that they’re voting for every 4 years or so. Inclusiveness and empowerment are probably the two most important items in Newark, New Jersey under Mayor Ras Baraka. So how does that translate into the technology space? We want technology that will empower residents to do better things with their day-to-day lives and with their long-term futures. We talked about land use a moment ago, but I think also just general services from the city, whether it’s health-related services or keeping the city beautiful and clean. Those are important things. So what do residents want? I think they want to be empowered in what they really want to be doing. No one really wants to spend their whole day at city hall trying to get a bill resolved or get proof of a certificate of occupancy from the permitting office. If we can empower people to fix up their houses faster by reducing the time it takes to get a permit and making it really easy, let them pay online, let them take care of it, be done with it, they can do what they want to do, which is expand that bedroom and plan for a family or rent out a property and make some extra cash. So I think it’s about empowerment and inclusiveness.

Sanjog Aul [00:19:37]:

So while you speak about that, do you see any specific social, legal, and ethical implications if the same work is being done and is being impacted by AI? And as per Doug, you don’t want a humanized machine. At least you want to envision it while you’re trying to plan for a family that something like that is impacting you.

Seth Wainer [00:19:57]:

I would think generally, along Doug’s point, privacy is probably the most interesting nexus to focus on because it can be policy driven. I think generally the law is pretty far behind on some of this stuff. Just thinking about the Open Public Records Act, for example, the burden of producing things on paper and printing them out and all these types of things that were relevant 50 years ago when that was a government record. To Doug’s point, your water bill is public record, and permits that have been pulled on property are public record. Using our official intelligence to make these processes faster while respecting people’s privacy and efficiently knowing who they are is important. Know your customer is a known item in the finance world. I think government could do well to learn from that and say what are the parameters we want to make sure that this is Seth and he’s authorized to do this, this, and that. Maybe do that once and have some sort of process, use AI to make this more streamlined. Right now, if you’re me and you go to the health department, you have to prove who you are. Then you go to the zoning department, you have to prove who you are. You go to economic development for rent control purposes, you have to prove who you are. There’s a lot of proving who you are without any strong continuity there. So I think software and artificial intelligence might have a role to play in that space.

Sanjog Aul [00:21:29]:

So let’s dig deeper on the privacy side. So Doug, what’s the flavor of privacy which is causing concern when you bring AI into the mix?

Doug McCollough [00:21:38]:

I’ll give you an example here. To your other question, what are we selling and what do people really want, I think the good question to ask is, hey, government, why can’t you do that? Consider what people’s experiences are and even ours when we go home from work after we work in a local government. You can speak into your phone and get a Yelp app and say, where’s a good place for me to eat lunch? It will give you a few results and rankings; it will say this place is terrible because of this reason, and this place is great because of that reason. It’s not particularly fair, but that’s okay because society has accepted that these results may or may not be completely accurate, and I can make basic decisions about that. So hey, government, why can’t you do that? We have a different burden. If we put data out, it has to have the imprint of government. Our concern is trying to be as fast as the latest app a private company might put out while being accurate enough that a person can rely on their data being completely correct. We can’t be just really good or really interesting and fast. We have to be completely and totally accurate. That’s a burden the public is not prepared to give up yet. In some of the examples given in terms of a permit or an identity or a dollar amount on a bill, these have to be completely accurate. The records shown from them have to be official. That’s a real challenge for us. People expect what they receive to be as fast and simple as Yelp or Uber, but they also expect them to be imprinted and official so they could show them to the police or a bill collector showing up at their door. It is a record, and that’s a challenge.

Sanjog Aul [00:23:38]:

Now, Seth, if I were to come to you and say, okay, AI when you heard it first, since being technologists, you might have heard it many years before a regular citizen would. But since it is evolving, I would not say we cannot trust AI at all, but it is still a long way to go. Government usually is not the first mover, and it shouldn’t be because it has got the citizen’s safety and well-being at stake. So should you even go ahead with AI or a sliver of AI which looks like it’s fully cooked? What’s the approach?

Seth Wainer [00:24:13]:

I think we’re not talking about using AI to make critical decisions that a human should be making. Without discussing the military approach, just looking at the civic and municipal space, I think there are things we are not doing that AI would help us do faster and cheaper. One example Doug brought up that’s relevant to us is cybersecurity. If there are AI tools that can better look for correlations and aberrations on the internal network, that will be extremely cost effective because a full-time analyst who’s literally just looking through log files is a very expensive salary, and the individual is only going to find so much. So it’s a good example of something we’re not doing right now that AI would help us address. We’re taking steps down that road as the software world starts to put out meaningful products that will fill that void. I think we have a ways to go with a lot of low-hanging fruit before we get to ethical decisions about whether the software is making the right decision. Even without a 311 shop or help desk, most of the phone calls I get to my team are, my computer is not working or my email’s not working or something like that. I think even if the AI world can improve on that basic customer service stuff without annoying people, then it’s going to have a lot of uses before we get to questions of morality.

Sanjog Aul [00:26:10]:

So, Doug, in your world or from your perspective, where should government stand in terms of AI adoption? Should it just go and do those safe, quote-unquote safe things and address those and then wait for something which is going to have a direct impact on a citizen’s life? Is that where we should draw a clear line?

Doug McCollough [00:26:35]:

In some senses, yes. We do have certain burdens in government, but we don’t have other burdens. In Dublin, Ohio we consider ourselves forward-looking and want to be advanced and an early adopter. Government should do what is safe. To Seth’s point, we are being guided by AI. We are not allowing it to make all of the decisions, but it can lead us toward a more efficient use of our resources. Citizens really like that. Public safety, crime prevention, and predictive analysis are valuable uses of some of this technology, and citizens love when you do that. You can direct police organizations toward a more efficient use of their time. That’s safe. We’re not going to make arrests or pull people over just on the basis of AI. We can innovate in ways that guide us toward more efficient decision making, test, and adopt new use cases and applications as the rest of the industry continues to advance.

Sanjog Aul [00:28:02]:

Based on the responses we discussed here, it looks like we are really not going to the citizens to ask their permission. We are essentially making policy decisions based on decades of experience dealing with the citizens to say what will allow people to stay safe and what they will not totally cringe about, and let’s do that. Is that a safer way of saying it — we will still charge ahead, but we will not alienate the very citizens we are serving?

Doug McCollough [00:28:38]:

I believe our citizens don’t want to have to tell us, and they don’t want to be asked. Our citizens expect us to constantly get more efficient, constantly look for cost savings, and constantly advance our technology while keeping us safe and not adopting anything that’s going to break anything. I do believe a constant interaction with citizens informing them of some advancements we’re doing is necessary. But we’re not necessarily asking permission or asking them to direct us. We should know that much. When you talk to local government constituents, they will tell you, what are you even asking me for? Of course you should be doing that.

Seth Wainer [00:29:27]:

And I’d like to add that. We’re asking citizens every 4 years. We’re asking voters every 4 years. Depending on the voting schedule, if you’re a council member or mayor or alderman or whatever the form of government is, we’re asking residents all the time.

Sanjog Aul [00:29:45]:

Let’s take a quick break, listeners. We’ll be right back. And let’s talk about if at all we have to make a business case because money doesn’t come easy. It’s a new technology, and it’s not just the technology where investment will have to be made. We’ll have to shift some processes, some human labor, and, of course, some funds. How do we go about it, and what are the challenges related to it? Please stay tuned. We’ll be right back.

Sanjog Aul [00:32:17]:

Welcome back. So, Seth, we always have to make a business case, and there are human lives attached. You’ve got funds and many other things. What is the approach that we are to take in making a business case for something like this? Yes. There would be small efficiency-related benefits, but still, we are spending money on something we haven’t spent earlier, and money is a scarce resource.

Seth Wainer [00:32:44]:

On the security front, for example, that’s very cost effective: here’s a piece of AI that’s going to help the city be safer rather than here’s a full-time analyst or a 24/7 contract for remediation plans and that kind of thing.

Sanjog Aul [00:33:04]:

And so, Doug, what about the fact that on one hand we are introducing this AI as a resource which is going to help reduce the amount of effort or maybe get things done faster. That looks like a great business case, but there is also a very active pushback because it may replace humans.

Doug McCollough [00:33:27]:

I think there’s a common belief out there that most people believe fewer humans in government is a good thing. I shouldn’t say most people, but people don’t cry for us. I’ll give the example of automated teller machines and banking and how that was going to get rid of all the people and branches and everything would be through a machine. What we found is banks are opening more branches and have changed their services away from counting money to more direct personal connections. In my dream of dreams, I would see a future in which people could interact with their government on a much more personal level. Instead of pushing papers around and making copies and stamping things, people would be more engaged in pushing out toward the community, learning what’s happening with people, describing and explaining things, and having a more personal touch. I am not convinced although I am a pessimist and think we have a lot to worry about  that adding AI to some of the things we’re struggling with as humans, such as analyzing security patterns, is going to necessarily reduce the number of people applied to the kinds of services people really want their government to do. It is a change and a disruption, and the work may be very different, but not necessarily less.

Sanjog Aul [00:35:09]:

So interestingly, when you look from a taxpayer standpoint, they’d love you for your response. But someone who is within the government organization who is tasked with something and tomorrow you say that, okay, I have been able to optimize this effort and thus it has been taken off your plate, and we keep incrementally taking it off their plate to a point where they are no longer needed, then they might see the whole progression differently.

Doug McCollough [00:35:41]:

Totally agree. I’ll just give the example of that security analyst whose job we just discussed and said a machine can analyze things faster and better. That is not a trivial issue, and I think that’s the challenge. Deploying AI is not that difficult; it really manifests on a machine. It operates on a server or in the cloud and over a wire. Deploying it within an organization is where the real challenge is.

Sanjog Aul [00:36:10]:

So, Seth, let’s talk about the readiness check. Do you think any government is just slapping this on as a tool, or could this become the fabric? Like earlier technology, of course, we are using it for certain point problem solutions. Similarly, AI could actually be brought into almost every aspect of government functioning. So before you lay a fabric, you have to look at the foundation. What are we looking at today so we know we are best suited and best ready for AI?

Seth Wainer [00:36:39]:

I think it depends heavily on where the market ends up going. The private sector stuff I’ve seen on AI is mostly around voice recognition. That’s the stuff that’s now in my living room with Alexa and on cell phones with Siri. So first touch point customer service seems to be where most innovation and research dollars are going today. Ideally, some of the higher-end analytics like the types of things Palantir would offer, to do fraud detection or those types of analyses, can also be scaled down to the municipal level. I think what municipalities will have on their plate is mostly dependent on where the private sector drives this technology. I’d like there to be AI around problem solving for IT problems, where there’s a desktop person and if anyone can’t get on the Internet, they click an icon and have a conversation about what’s going on. We analyze the computer: are there proxy settings like this? Is a cable unplugged? Have they logged in correctly? All that stuff. But I haven’t seen the private sector move sufficiently to make that cost effective. You have to adopt only what’s really cost effective. We’re not going to spend millions to put in crazy might-work systems.

Sanjog Aul [00:38:15]:

So one thing is, of course, you’re saying where as compared to others, where we will get and eventually over time we will be at the right level of adoption and the way we use AI. That’s great. Now before you even get going and build a foundation, what would we be doing to check to see whether our government, the processes, and other things are at a point where we can start inviting AI, not just superficially, but actually embedding it in the DNA?

Seth Wainer [00:38:47]:

That’s a valuable question because in a software world, even non-AI solutions, the number one problem I run into when looking at a process is that it’s not a determinant process. It’s not like 1, 2, 3, 4, 5, it’s a whole bunch of confusion and humans involved that make it effective for the applicant but difficult to model. I don’t think AI is going to be all that effective unless governments have perfect processes in place on whatever it is, whether it’s applying for a new business or a road opening permit. We’re only going to be able to use technology to improve the process if the process is really strong, tested, and binary. It’s either working or it’s not, not, I talked to Janice and Janice got me that stamp real fast because Janice is on it. For an expedited payment, bam, it’s done. So to get ready, we have to get our own houses in order. I think that’s the most important thing.

Sanjog Aul [00:39:58]:

So, Doug, in your word, the way you see the internal processes today, and yes, Seth mentioned that we have to perfect the processes, but there’s nothing like perfection. We just make progress towards it. What role would you put in place? Even before that, if you have a checklist to say, do I have this, this, and this? Then I will go ahead and invite AI. Do we wait for that perfection before we start inviting, or do we do both in tandem?

Doug McCollough [00:40:25]:

I think we do both in tandem. I don’t have a checklist; I don’t think our industry does. I think the closest thing to it is the developmental steps toward AI. One of the big ones is automation, which we said is not AI. Then data analytics — these are things you do before you have AI. Do you have your data in order? Do you have your decision making? Can you even do that without a machine? Can we automate? Then an AI can move swiftly across business units. I’ve often said a city is a small state; it has all the agencies and departments just in one organization. I think AI will creep in. It is creeping in. We will trip over it and find it’s already there. Many systems our departments use — tax, finance, police, fleet management, parks and recreation — are already using systems on an operational level, and those systems are being updated with AI in the cloud. It’s not necessarily a decision a CIO of a city made. It’s an operational improvement somewhere off-site. We’ll find AI already in our environment more so than internal city operations at an administration level. That’s probably the last thing to go. We will work on automation and data analytics and open data, and AI will find itself there in a much easier way because we prepared the steps.

Sanjog Aul [00:42:30]:

As a taxpayer, I personally would not like any leakage in the sense of people making decisions adopting policies to my best interest but not in the most effective or efficient manner. If we naturally allow AI to creep in, which it is today, and it gets slapped on a not-as-efficient process where it will not create the most value, well, nobody’s going to come and fire someone, but it is still leakage. So in all honesty, when people are within the organization or as a citizen sitting outside and watching, I would like it to go through a critical path versus somehow getting there. If we take a solemn vow to say all citizens and all officials of government are going to invite AI but will do it in the most effective manner so leakage is minimized or eliminated, what would that look like? If nobody’s been asked today whether there is leakage or not, but if we take it in a cavalier fashion, AI might cause leakage or we might be losing taxpayer dollars or not using it in the best way possible. So what would that tight approach look like? Please stay tuned, listeners. We’ll be right back and discuss.

Sanjog Aul [00:46:04]:

Welcome back. So, Doug, your response was regarding bringing AI in as a natural fashion like any other technology. My challenge was to not do it so that we create leakage or encourage leakage. What could we be doing in such a manner that the leakage is minimized or eliminated?

Doug McCollough [00:46:32]:

My description of how AI and other technologies may enter our environment as sort of just happening without our control may have seemed a bit cavalier, and maybe it is. That can be very scary with something like AI. I would say there’s a number of hot-button areas of people’s lives that they are really interested in what technologies are planned to be used and how. We need to focus on communicating around those. We should create forms of communication in which we can interact with members of the public — taxpayers who want to know how does technology get into my city? How do you adopt it? These are processes many people have not concerned themselves with in the past. Because of the advancement of technologies and people’s sophistication around them, they’re going to want to know. We’ve had a lot of attention around open data, but most people don’t really understand that technology; they just want to see the data. We can answer that by exposing to people how we are going to make decisions about what comes in. I would draw a distinction between a sleep management application or a parks and recreation application that may begin doing AI versus the core identity service decisions that we make as a city around a table. We can do that, and I should be careful about how I communicate about it.

Sanjog Aul [00:48:09]:

So, Seth, when you look at the most optimal way of bringing AI in, would you take a critical path approach where you would have cleaned up the shop and then brought it in? Is that a possibility, or is it just pie in the sky?

Seth Wainer [00:48:22]:

Well, I think that that’s, It’s an opportunity, right? You are at an inflection point. Everything’s an opportunity. The idea that AI poses more of a challenge or is under greater pressure to be perfect than other systems like cell phone networks or radio networks or even upgrading to Windows 10 these are all things public sector technologists have to decide how they want to approach. When we assume our processes are perfect and software somehow will make it correct, that’s pie in the sky. The idea that any one thing we’re doing is absolutely perfect and we really can’t improve on it is where we’re lying to ourselves more than the AI side. The important thing is to think about how technology can be deployed cost effectively to impact residents and empower them to do what they want to do. If it’s going to help them get a permit, zoning is a good example. Residents want to do things but don’t know if they’re zoned appropriately. If we had an AI that could determine how many bedrooms it is, do you have proof of documentation, have you been living there, are you the owner, then you can do this or that if that could be done through AI with accuracy equal to what a person could do, I don’t think any process we’re doing is absolutely perfect and flawless. If the software is good and you can prove it, I don’t think the software is going to cause anything worse to happen. The question is how you’re deploying it and where you’re focusing on pain points. I wouldn’t worry about it.

Sanjog Aul [00:50:24]:

So would you say that you’re calling AI essentially a tool, an aid?

Seth Wainer [00:50:31]:

Yeah. It’s the same as talking over a VoIP phone. When you change over from copper phones to VoIP phones, it’s a tool. The question is to what extent is it going to address the needs of the employee, how it’s going to save cost, and is it relevant? Inevitably, there are hiccups, like having to dial 91. But the cost savings are tremendous. The same is true with software and AI: if you’re going to use it to find policing hot spots, you want to identify those hot spots rapidly. Maybe you save money on analysts. Analysts can focus on other parts of the investigation aside from identifying the corners. They’re not just looking at 30 days’ worth of data; they can look at 60 years worth of data, abandoned property history as it relates to that hot spot. AI is a strong tool in the quiver for how we’re going to make government more effective. 

Sanjog Aul [00:51:36]:

So, Doug, do you see or foresee processes and policies being altered where not just the AI portion, but humans, robots, algorithms, and other things are going to coexist because AI is actually kind of the backbone for all of this?

Doug McCollough [00:51:55]:

I totally agree with Seth on this. That’s the point I was trying to make about incremental change. Across 10 years we’re going to see major policy change, but I don’t know that we’ll see that in one or two years. Using a different tool set doesn’t necessarily require a different policy change, but it probably requires a different culture and training. People will use tools differently and interact differently, requiring us to work differently together. Our policies are not so rigid and inflexible that they can’t adopt new ways of looking at the world. If you look at AI as a black box, it’s better and more efficient. It’s scary, but it’s not going to come all at once. Even the data we put into it and the tasks we ask it to do are incremental. I don’t think anyone is turning over an entire police department or its analytics to this. We are beginning to use it more and more as a tool, and our policies will change incrementally. We’re thinking a lot about connected vehicles and autonomous vehicles, which will require law and policy changes, but we’re years away from that and can make the changes incrementally.

Sanjog Aul [00:53:29]:

On behalf of the show and our listeners, thank you so much, Doug and Seth for sharing your views on how government can leverage AI, which is artificial intelligence, to move forward its agenda to provide a progressive and secure environment to its citizens and constituents. Thank you so much.

Doug McCollough [00:53:46]:

Thank you.

Seth Wainer [00:53:48]:

Thank you for having us.

Sanjog Aul [00:53:50]:

And, listeners, please like us on Facebook. Search for CTN, and be sure to follow us on Twitter and join our LinkedIn community. Thank you again for listening to this segment on CTN. This is Sanjog Aul, your talk show host. Till next week. Take care, and God bless.