--- id: "workshop:beyond-displacement-building-an-ai-ready-workforce" type: "InnovateUS workshop" title: "Beyond Displacement: Building an AI-Ready Workforce" url: "https://innovate-us.org/beyond-displacement-building-an-ai-ready-workforce" date: "2026-06-11" duration: "60 minutes" language: "English" series: "Elephants in the AI Room" instructors: ["Rob Asaro Angelo","Vanitha Zacharias","Katrina Flory"] moderators: ["Deborah Stine"] recording: "https://www.youtube.com/watch?v=3RFNVusazHQ" transcript: "automatic, from the recording" source: "InnovateUS (innovate-us.org)" licence: "CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/), as stated on innovate-us.org" --- # Beyond Displacement: Building an AI-Ready Workforce InnovateUS workshop · 11 June 2026 · 60 minutes AI is already changing how public employees draft emails and documents, analyze, communicate, and deliver services. For governments, the workforce question is not only whether AI will replace jobs, but how agencies can shape adoption so that workers are heard, and receive appropriate training so they are able to focus more on the mission that brought them to public service in the first place, rather than looking over forms, copying and pasting from spreadsheets, and data entry. This session explores the workforce impacts of AI in the public sector, including displacement risks, changing roles, productivity pressures, and opportunities to make public work more enjoyable, effective and sustainable. Participants will examine strategies for workforce planning, worker engagement, training, and policy design that help agencies prepare for AI while protecting dignity, accountability, and service quality. This workshop is designed for public professionals involved in workforce planning, operations, labor relations, service delivery, AI adoption, human resources, and organizational change. ## Learning goals By the end of this session, participants will be able to: - Describe how AI and automation may affect public sector roles, tasks, and working conditions. - Identify opportunities for AI to improve public work while recognizing risks related to displacement, deskilling, surveillance, and productivity pressure. - Apply strategies for engaging workers in AI adoption, including consultation, training, feedback loops, and shared governance. - Evaluate policy and management tools that can help agencies strengthen worker power, build AI-ready skills, and protect public service quality. ## Speakers - **Rob Asaro Angelo**, Senior Fellow, Burnes Center for Social Change & Professor, Northeastern University - **Vanitha Zacharias**, Chief AI Strategist and Head of Technology Governance, Department of Administrative Services, State of Ohio - **Katrina Flory**, State Chief Information Officer, State of Ohio ## Moderators - **Deborah Stine**, Senior Fellow, Innovate US Part of the series [Elephants in the AI Room](https://innovate-us.org/workshop-series/elephants-in-the-ai-room). Recording: https://www.youtube.com/watch?v=3RFNVusazHQ ## Transcript Automatic transcript of the recording: names and terms can be misheard, and the times can be a few seconds off. Each time links to that moment of the video. **Deborah Stine** [0:00](https://youtu.be/3RFNVusazHQ?t=0) So, hello everyone. My name is Dr. Deborah Stein. I am a Senior Fellow for Innovate US, and I'd like to welcome you to the first session of our Elephants in the AI Room series. Joining us from Innovate US, behind the scenes there, is Christine N.A., and she is going to help you with any technical issues, any other questions you have. She'll be posting lots of interesting links and information, including the speaker bios, in chat. So, please feel free to reach out to Christine if you need help. Speaking of which, Christine, can you show the Elephants in the AI Room series so people can see the whole thing? And, oh yes, we have this poll that we would like everybody to fill out. It's a pre-workshop poll, just kind of telling us a little bit more about you and why you are here. So, the Elephants in the AI Room series is something, I think, that we all think about. As somebody, myself, who uses a lot of AI, teaches about it, I know I have somewhat of a guilty conscience sometimes, thinking about all the implications of AI for society. So, the purpose of this series, Elephants in the AI Room, is to address these most frequently requested and most difficult questions about responsible AI. This session really came from all of you, all the participants who were interested in this topic. As I was teaching how to do everything from prompting to 10 things about data centers, these were the issues that have come up over and over again. So, today is the first of these series. So, today we're talking about Beyond Displacement, about building an AI workforce. And then next week, on the 23rd, we'll be talking about energy and the environment. And on the 25th, we're going to be talking about [2:05](https://youtu.be/3RFNVusazHQ?t=125) people who have invisible barriers, disability rights, and public sector AI. So, we hope you will attend all of this series. It's an important one to do, regardless of what field you work in, to understand all these different things happening in the AI world. So, our speakers today are Rob Azzaro-Angelo, who is a Senior Fellow at the Burns Center for Social Change and a Professor at Northeastern University. Katrina Flory, who is the State Chief Information Officer for the state of Ohio. Vinita Zacarias, Chief AI Strategist and Head of Technology Governance for the Department of Administrative Services in the state of Ohio. We'll start with them in just a few minutes, but I wanted to go through a few housekeeping details. First, this session is recorded and the recording will be available tomorrow on our website. Sorry, the poll keeps popping up here. And we would also like to encourage you to rename and introduce yourselves in the chat. We'd love to know your state, your agency, your role. We get people from all over the world here, and we'd love to hear what you're doing. [3:31](https://youtu.be/3RFNVusazHQ?t=211) So, that's always good to know. Please use a name you would like on your certificate of attendance. A short poll is going to pop up on your screen, and we want to get a sense also in the room of who's with us. So, please make sure you fill out that poll. The other thing is we have lots of experts from different backgrounds in chat. So, if you have questions, you know, feel free to put those in chat. But for the official questions, you really need to use the Q&A function. And that we'll look at at the very end of the session for our speakers. So, we're going to start, each of them is going to make a brief presentation. I'll ask a few questions, and then I'll start answering questions in the chat. Sorry, answering questions from the Q&A section. And just make sure if you want them to have a chance of being answered in the time we have available, that you put them in the Q&A section. Christine is putting instructions in the chat now on how to turn on closed captions and also translation. With that, we're going to go to our first speaker. Rob, that's you. And let's go ahead and get started. **Rob Asaro Angelo** [4:39](https://youtu.be/3RFNVusazHQ?t=279) Good afternoon. Good morning, everybody. Depending on where you are, I'm still thrilled to be here with all of you today. Like Debra said, I'm Rob Rosario-Angelo. I'm the former Commissioner of Labor in New Jersey for the past eight years, up until January this past year. I'm now a Senior Fellow at the Burns Center for Social Change. And I love doing these programs, mainly because I get to see so many different public servants from around the country at different levels, different titles. So, everyone's popping their name in the chat. I get really excited to see who I'm on here with today. Let's see if I can figure out how to share my screen. I practiced, but who knows? Let's see. Boom. Can you give me a thumbs up, Debra? I will. Success. Great. And you may or may not know, I talk really fast, so I have notes here to help myself slow down. So, here's what we're here for today. So, first, I love the title of this series, Elephants in the AI Room. So, first, I wanted to, oh, come on, work now. There we go. I wanted to show you my favorite elephant. Lucy the Elephant, a historic landmark in Margate, New Jersey. Became famous a little bit during the National Community Vacation. On the opening credits, you get this little postcard of Lucy. And amazingly enough, on the left-bottom side there is the inside of Lucy, which started as a hotel back in the late 1800s. And it's still alive, well and kicking, in Margate, New Jersey, and a great tourist stop for anybody. And I go there multiple times a year whenever kids are near me. [6:05](https://youtu.be/3RFNVusazHQ?t=365) So, I want to talk about some of my background about why I'm passionate about this work. During COVID, in New Jersey, our unemployment system, like most states, buckled. Not technologically, but just the overwhelming amount of claimants and working on a system, trying to come into a system that was not designed for a COVID emergency, much less people's everyday emergency when they are displaced from work. But it shaped everything I did afterward. Relentless focus on reducing administrative burden for residents, and importantly for me, for our employees. And a lot of our folks in government, especially in the human services area, are now being pressure tested with H.R.1, the One Big Beautiful Bill Act, and all the new work requirements, all the new changes that are in there. So, I've been working a lot with my former colleagues in human services about that as well. Okay, so here's a quick timeline of our AI journey in New Jersey. Our first department, all staff AI training. We're super proud to be the first department to have every member of our staff participate in the training. It was in 2024. We also launched in New Jersey that year, NNJA AI Assistant. [7:11](https://youtu.be/3RFNVusazHQ?t=431) I used the AI sandbox a lot, and I also tried to lead by example. When I emailed out to our staff about getting trained on AI, I showed them how I used it, how it was helping them, helping their colleagues in day-to-day grind of working for a labor department across the board throughout the state. That wasn't just something that was coming down from on top. It's something we wanted folks to use to make their lives better, quite frankly. We used AI on a big language access project with the U.S. Digital Service and others to make sure we had better translations for the multiple populations that we serve. We serve many, many languages in New Jersey. We're an extremely diverse state. Making sure we had not just culturally appropriate, but accurate translations for not just every language out there, but the different dialects of language, different countries of origin, and everything else. We were very proud to win an award on that on the right side there, the U.S. DRC award for our work on language access, which I think is one of the, quite frankly, easiest ways to use AI in your system. I don't think anything's easy in government, but something that could have instant improvement for an important population. [8:16](https://youtu.be/3RFNVusazHQ?t=496) Underlying all this and why I brought some of that stuff up is you can have the best methods and procedures. You can come to all of our trainings. You can scribble down every word we say today, but I think we had success because our relationship was built on trust. If your employees don't trust you or don't think you have their best interests at heart, it's going to be tough to roll out any new thing, whether it be AI or any other new product. From day one, the New Jersey Department of Labor, based on my time in the Obama administration, we focused very strongly on giving our workers a stronger voice in the workplace, increasing their benefits and pay, making their work-life balance better, increasing tuition reimbursement. Just really from day one, I wanted to make sure that everybody was on board, part of the same team, that we were all involved in decision-making. I think that sort of shown through and was successful. I really account a lot of our success during COVID and afterwards to that trust we had built up and down the Department of Labor. [9:18](https://youtu.be/3RFNVusazHQ?t=558) I'm super proud. One of the things on here is our internship program. I saw that one of our interns is on this session today. Hi, Prakriti, one of our great interns. We have like 60 interns every summer after our third summer through, and I'm really proud of that program. It's been a great recruitment and retention program for us as well and given us great, amazing staff throughout the department, which is another way to reduce the administrative burden of your current staff is to get them more help. Our program has been great to not just get more help, but to provide a career pathway for really smart folks in New Jersey. I just wanted to put that back there. I know that sometimes whoever's on this call might not be in charge of some of these things, but it is still the kind of thing you could try to fight for or talk about with your staff. Clearly, we had a very good boss, my boss, Governor Murphy, that allowed us to do all these things to help our staff, frontline staff, bargaining staff, management staff, all the way through. Some of the things I think that are important and some of the things why you're here today, and one of the reasons why we've been such a great partner, I'm sorry, my former self at the Department of Labor was a very strong partner with Innovate US. [10:22](https://youtu.be/3RFNVusazHQ?t=622) I think we were the first state to roll out their training. Back in the day, I'm on AI, responsible use of AI. We want to make sure that our folks had a basis of knowledge, but also a sense of security about the work that we were going to be undertaking. We made the training mandatory, like you have mandatory trainings for ethics or whatever else you have in your state learning system. It was one of the best received trainings that we put out there in the state. I really think it came at a time where folks were just really curious about it, and it gave folks a way to sort of twit their beak to get ready for this upcoming AI work that we're doing. Importantly, in all of our work with technology or anything else, quite frankly, start with the mission, not the technology. The technology, AI, or anything else should be there to solve a problem, language barriers, backlogs, administrative burden. We never deployed AI or any technology just to have the new technology. We wanted to make sure that we were something that was being useful, and I'm very proud of the technology that we rolled out, whether it be our unemployment modernization in the department, which I think is world class and seen by the states as a model. [11:28](https://youtu.be/3RFNVusazHQ?t=688) We were the first state to use drones in our wage and hour and public safety enforcement areas, and many other examples of how we were sort of pushing the envelope on to new technologies, and I think our staff like being part of that. They like being part of something new. Breaking news, sometimes the government, the work is repetitive, and you get monotonous. So I think anytime you're introducing something new and exciting is a way to motivate your staff, a way to get them interested in the mission that we're talking about. And as long as you're staying focused on the mission. And I think you've heard me say throughout, sorry, there's a floating four there. I got rid of one of these things. Workers are our partners, never our obstacles. And I think that's what I was talking about, that trust before, that our folks knew that from day one. Certainly, when we went through COVID together, and I'm sure many of you on here in government have gone through COVID with your staff, and while departments of labor were certainly overwhelmed, everybody who worked in government at that point in time was sort of in the foxhole with their colleagues. We served our co-workers, we served our residents. So I think that just inherently built trust. [12:29](https://youtu.be/3RFNVusazHQ?t=749) And because of that, we were forced into partnership, not that we weren't already there. But when you're there every single day, answering to 2 million claimants instead of 20,000, or whatever you were doing with all the black swan events everywhere in every government service, you just by nature, I think you have a stronger camaraderie and hopefully a better trust level as well. When I talk about our workers, I don't mean just the ones that work for us, but also the ones that we're serving. We're the department of workers, essentially. People come to us for workforce training when they're having their wages stolen, or their overtime not paid, or they're injured, or they're caring for a sick family member, or they're put out of work through an injury, or they're unemployed. Everything we do in the department was always about the workers. So we had to make sure that we were getting their feedback, because they weren't just those that we served. They were our partners as well. They were our eyes and ears on the ground, and they were the ones that sort of helped us find problems before it got too widespread. So always making sure that we're involving them. I don't think I need to talk too much about the feedback loop between the folks that we serve and others. [13:34](https://youtu.be/3RFNVusazHQ?t=814) Oh, did I miss a slide here? Oh, okay. So the opportunity ahead. Just real quick, on the left side are some of the things that we did launch in my time as commissioner that we're very proud of in the AI world. One of the things that we used was actually mycareer.nj.gov, which is a personalized career pathway tool for New Jersey workers, matching skills opportunities statewide. And as a state labor agency, we're very lucky that we have all the wage records for the state. So it wasn't just based on hypothetical data or scraping LinkedIn accounts or something else. We knew what every worker in New Jersey has done for their work and how much they earned over many years. So we're able to use that data to feed our system to give, to show folks a realistic career progression, wage progression, everything else based on real data of their coworkers in New Jersey. We launched AIDA, which is a UI appeals tool, which lets our, when someone files an appeal for their unemployment benefits, they have to go to an appellate appeals hearing. [14:34](https://youtu.be/3RFNVusazHQ?t=874) And at the end of the hearing, the appeal examiner has to write a decision essentially. But that involves a lot of research, looking at the proper judicial sites, presidential sites, UI regulation and statute. So what the tool does is that as well as the hearing is going on, it is pre-populating the background for all those things we just talked about. The legal research, all these things that just take time after a hearing is over. In the end, if the appellate specialist was verifying everything that was said, So here are some of the risks that you have to manage proactively. Again, I think this applies to all technology, but certainly with AI, just because the speed of it. Displacement anxiety is real, even when jobs are just redesigned, not eliminated. We haven't eliminated any jobs, and I'm a firm believer that AI is going to lead to more jobs in government, quite frankly. [15:37](https://youtu.be/3RFNVusazHQ?t=937) Deskilling, when AI does too much, eroding expertise over time, that is a real concern. We are now, when we're using it successfully, we're handing over some of the monotonous, repetitive work so we can focus on more humans, but at the same time, you still need to know how that process works. If you lose power or whatever kind of coop planning you have to do, somebody has to know where to put the paper that you're now uploading into your internal AI system, how that interacts with the benefit program, what the eligibility determinations are. So just making sure that as we're using this technology, we don't lose the institutional knowledge that is now being transferred over. Equity gaps, make sure that we're testing all of our tools with a varied set of workers, whether it be by age, by length of service, with the department, where they worked in different parts of agencies before, so making sure that everybody has the same comfort level with their user experience and their knowledge of AI or, like I said, any other technology. [16:39](https://youtu.be/3RFNVusazHQ?t=999) Accountability gaps, AI touches, benefit eligibility decisions, again, we didn't really have that so much, but something has to be addressed up front to make clear to folks that that is not, one, the reality of what's going on, you need to make sure you have good processes. Luckily, at the Department of Labor, I think, I can't remember if I talked about this recently, during COVID especially, there was a big push by some states to be able to contract out more of the work to private third parties so they can make unemployment determinations on eligibility. USDL said, no, it has to be merit state staff, which was fine with us, we had no problems with that, but some states did. But importantly, that showed us that even all the new things we were doing during COVID, putting in new programs, new technological fixes there as well, that in the end, every benefit determination had to be done by a human. This was before AI, just about, it couldn't be some third party. So we were fully ready for that kind of dynamic. During the pandemic, we did have a contracted call center, but that was only sort of a frontline, answer basic questions, and then things were elevated to our actual state UI employees. [17:42](https://youtu.be/3RFNVusazHQ?t=1062) In a way, that was sort of like an AI filtering technique, these files would come to us with codes from the private call center about what the issue was, what kind of problem it was, and our real staff would have to call them back or make some kind of decision. So we sort of had this kind of loop already, where our own agents knew that there was some other tool, in this case, other humans that weren't state employees, that we had to make the actual determination on. So sort of my takeaways for you today, and I'd love to answer questions later on, that's sort of my favorite part of these webinars. AI is not going to save public agencies from hard policy choices, but use well, it can help dedicated servants do more what brought them to this work, and less of what burns them out. What I always say is that I think AI should be used, and when it's used successfully, in government specifically, that AI gives our humans more time to talk to other humans. And I believe in that strongly, whether it be a claim specialist, we're always overwhelmed on the phones and everything else, whether it be our wage hour investigators, who have a case backlog, our disability employment specialists, anybody who's got a desk with a bunch of paper on it, or cases, or a big folder of things they need to work on, anything that can get them away from that, and be out there in the community talking to folks, calling back workers, letting them know what their problem, what they might need for them to get their benefits moving forward, all of it. [19:02](https://youtu.be/3RFNVusazHQ?t=1142) I see AI as a tool to give us more human interaction. So train first, be honest and open about the pain that might be coming with it, build with, not for, make sure your frontline workers and community partners are involved, and also making sure that you have government frameworks that can catch up the technology. I don't need to talk to folks here about how long government processes are, I don't mean to focus on procurement, but it could be hiring, it could be difficult to hire folks who know about AI, because of state hiring rules, or whatever it may be. So just make sure that the infrastructure around is ready to contemplate all these changes. The last thing I would say also, and I talk positively about using AI to help our government workers talk to more humans, but let's be honest, a lot of our government workers are in their jobs because they don't want to talk to humans. They like going through processes, they like looking at databases and switching the data over, whatever it may be, and that's totally fine. So you've got to make sure that those are the folks that feel the most comfortable, quite frankly, the ones who don't want to have more human interaction, but there's plenty of work to do in government. [20:04](https://youtu.be/3RFNVusazHQ?t=1204) That's why I am less worried about displacement in the public sector. We've been understaffed pretty much since the invention of government. So I always felt strongly that it's only going to make us, not just be more effective, because I'm a little worried about the framing that AI is only about making us be more productive and more effective and have better metrics and deliver more services. It's going to make us happier workers. When I talk to folks about how they're going to use AI in their work, I say, what is the thing you hate most about your job or what thing do you dread when you get to the office that day? Usually, I think it's something that an AI tool can help with. So I'll stop it there. I look forward to hearing the presentations from the others as well. Thank you so much for being here today. **Deborah Stine** [20:40](https://youtu.be/3RFNVusazHQ?t=1240) I will stop the show. Thanks, Rob. Great to hear from you. And now let's go to Vinita and Katrina. All right. **Katrina Flory** [20:55](https://youtu.be/3RFNVusazHQ?t=1255) Thanks, Deborah. And that was great, Rob. Appreciate hearing everything that's happening in your state as well. And I think we'll find that we have a lot in common in our approaches. Vinita and I kind of discussed a couple of things as well. So happy to be here today and see so many folks in the chat from across the country. So that's pretty exciting. **Vanitha Zacharias** [21:19](https://youtu.be/3RFNVusazHQ?t=1279) Thank you. And I'm happy to be here. I'm from the state of Ohio also, working in the state CIO's office with Katrina. And I'm seeing a lot of folks from Ohio as well. So glad to see all of you here. Thank you. You're on mute, Deborah. **Katrina Flory** [21:41](https://youtu.be/3RFNVusazHQ?t=1301) Yeah, Deborah. **Deborah Stine** [21:43](https://youtu.be/3RFNVusazHQ?t=1303) Sorry. So Katrina, let's start with you. Yeah. How do you evaluate policy and management tools that can help agencies strengthen the public workforce, build AI-ready skills, and protect public service quality? **Katrina Flory** [22:00](https://youtu.be/3RFNVusazHQ?t=1320) Thanks, Deborah. So AI-ready is in the title of this session. And I think that was something that our AI Council was focusing on from the get-go. We had our policy out. And then that focus was really on our workforce and how to make sure that they are ready. And we kind of went with a tiered approach. So foundational training, partnering with Innovate US for responsible training for the public sector workforce. And we talk about this almost every conversation we're in and everywhere we go, the importance of that foundational training. We customized it with the information from our policy, IT17. And we also developed a lab so that as folks are trained, they can go into the lab environment with those tools. So that foundational training is very important. I think we've had over 15,000 of our employees take that training. And then we just recently rolled out the responsible AI training for legal professionals. We actually work with the Supreme Court of Ohio to have that have CLE credits as well. [23:01](https://youtu.be/3RFNVusazHQ?t=1381) So that's rolling out to our legal staff. And then just looking at those various verticals where some of those specific roles can leverage some of that training to bring to their expertise, whether that's legal or IT or procurement, whatever that is. Because we know that this is not just an IT thing. We want to make sure there's awareness across the board with our staff. And so we also have our Learn Ohio, which is Skillsoft. And we create a role-based trainings in AI. So there are AI journeys there that our employees can go. And depending on what their role is, they can find appropriate AI-based training to use. We have everyday AI for folks. We've put out stuff around Copilot Chat, which they all have available through their 365 licenses. And then we've also created, like I mentioned, a lab environment with access to not only Microsoft tools, but CLAWD and Gemini, where employees can go and really work on those skills. And then in January of this year, we rolled out more of a specialized focus, what we're calling Thrive Thursdays. [24:07](https://youtu.be/3RFNVusazHQ?t=1447) So every Thursday, and actually right now, it's happening. We have a hands-on learning training experience for employees on-site and targeting specific technologies that are already in our stack. So we work with our vendor partners. So think of a Salesforce, a ServiceNow, Microsoft, all those folks. And we bring in their team. We have a hands-on experience for our team. And then when they go back to their office or their work location, they have access to those tools in the sandbox and lab environment to kind of continue that exploration and learning process. So we're really excited about those that have been very popular, and we've had a lot of folks there. And we specifically have made those on-site. We're trying to catch everybody where they are, right? So we want to bring people in the room, but we also record them so folks can participate in them later. We have online training that you can take at your leisure, those type of things. So making things available to folks has been very, very popular and a good approach, I think. [25:09](https://youtu.be/3RFNVusazHQ?t=1509) And then we also have various collaborative opportunities. We have monthly community of practice meetings. So agencies like our Department of Transportation is presenting use cases that they found benefit to their workforce or their programs. And the Department of Aging or Department of Natural Resources, any one of those, can take those and see how they could apply to a business problem or challenge that they're having. So we're trying to make sure that agencies are collaborating and sharing that expertise across the board. We also have engagement with our chief data office and their council. We have other communities of interest. And just those partnership opportunities have been very helpful to make sure that we cover all of our bases, if you will, and provide employees with those skilling opportunities that are necessary. So we're looking forward to that. Like I said, they've been very popular, and how we're promoting those and providing that clear guidance and opportunities for people. **Deborah Stine** [26:10](https://youtu.be/3RFNVusazHQ?t=1570) Great, thanks. And Vinita, how do you think AI and automation might affect public sector roles or tasks, their working conditions? **Vanitha Zacharias** [26:21](https://youtu.be/3RFNVusazHQ?t=1581) Thank you, Deborah. So we read in the news that AI is taking our jobs. And I think Rob also mentioned that this is a risk that we need to manage. So I want to put out a statistics there. Through Forrester's research, what they have found is AI and automation will cost only about 6.1% of total US job losses by 2030. So the key takeaway from this is we won't lose our jobs, like entire jobs, to AI. It will replace tasks within our job functions. And there are multiple layers to it. And let's peel the onion a little bit here. So let's take some jobs where there are a lot of routine tasks going on. An example of that would be a customer service tier one function or like a front office desk jobs. So there are a lot of routine work going on here. And those would have a lot of significant transformation. [27:23](https://youtu.be/3RFNVusazHQ?t=1643) Back in those days when computers were introduced, typing jobs eventually faded away with email and Word and all those tools. So those jobs would be a lot of tasks within those jobs could be automated. But then there is another layer of jobs like project manager jobs or software engineer jobs where a significant portion of that would be augmented through AI. An example of this would be a few years ago, project managers will be taking care of all the project meeting notes. And they would be taking care of action items. And all of this are done under 30 seconds after the meetings these days. And now they actually have more time to spend on addressing project risks and project issues. Similarly, software engineering jobs, like they can take on more projects now because AI is able to generate a lot of code and save significant amount of time. Now, there are like additional jobs like, you know, that those involve a lot of collaboration and co-creation and interdependent effort. [28:30](https://youtu.be/3RFNVusazHQ?t=1710) An example of those would be like a physical location, right? Like building maintenance jobs, they need to come physically to a place to fix certain things. Also like emergency management jobs where they have to coordinate with multiple folks. And also like highly complex legal analysis. These kind of jobs, AI can offer support, but it cannot, because there is not a lot of routine going on in there, it cannot replace a lot of those functions. And also, humans are required in these jobs to build trust. And Rob talked about trust as a prerequisite. It's very important to build trust and de-escalate conflicts. So I want to present another dimension here. So how can our workforce be AI-ready? So we need to focus on some of these key skills that only humans can bring, like critical thinking, adaptability, agility, and curiosity. Thinking of like those durable, long-term skills for success in AI-ready world. [29:32](https://youtu.be/3RFNVusazHQ?t=1772) And those would be cultivating growth mindset, human-centric communication, and then how to be comfortable in ambiguity, and also ability to make judgment under uncertainty. So these are some durable skills that we have to develop to create an AI-ready workforce. **Deborah Stine** [29:52](https://youtu.be/3RFNVusazHQ?t=1792) Okay, great. Thanks. And as I'm posing these questions, don't forget that the audience also has the ability. ...questions, just use that Q&A box. I see great questions coming in. Okay, so let's go back to you, Katrina. What do you think the opportunities are for AI to improve public work while recognizing risk related to displacement, desk skilling, I don't know what desk skilling is, so we can talk about what that is, surveillance and productivity pressures? Yeah, thanks, Debra. So, you know, **Katrina Flory** [30:28](https://youtu.be/3RFNVusazHQ?t=1828) we've been very out in front with our employees and our staff saying that, you know, we're not looking to replace anyone with AI, so we are looking to provide the tools to make your job better. Like Rob said, what's the thing that you hate the most? Maybe it's doing the wrap-up thing after a call, maybe AI captures that and all you have to do is kind of proofread it or something like that and you can get on with your day, but very careful around making sure that folks know that is the intent of these tools. Make your life better so you can provide better services more efficiently to Ohioans across the board. So, you know, I mentioned our AI Council and since that was established in January 24, we've approved close to 200 use cases for our state agencies, so the state agencies have to come with us to us with energy generative AI use case, so a lot of those are in coding, you know, creating training, supporting contact center workers, knowledge assistance, you know, gaining some insights from unstructured data, you know, leveraging it for cybersecurity purposes, grants management, operational things, you know, contract analysis, you know, our procurement folks have a lot on their plate, you know, when these massive RFPs come in and things like that and also fraud prevention. [31:40](https://youtu.be/3RFNVusazHQ?t=1900) There's been a big focus on that as well, but that's been more of our focus so that employees know what is happening in that space. We have a very strong labor union in the state of Ohio, OCSEA, and, you know, one of their concerns at the last contract renewal was specific to AI and how this was going to impact the workforce, so I meet with them on a monthly basis and share things that have come through the AI Council. These are tools that are approved that your folks are using or can use, you know, these are use cases that we're discussing, these are training opportunities that are available, and they've helped market those training opportunities as well. I send that information so that, you know, we try to get there from all angles and make sure our employees know that these things are available to them. Also, working with our statewide Office of Talent Management and Human Resources, that's another channel that we're sharing that information with just to make sure that we're being, you know, transparent in what we're doing, making sure that these things are available to our employees and that everyone understands that we are putting our culture first over technology. [32:50](https://youtu.be/3RFNVusazHQ?t=1970) We're trying to put these tools in their hands so that they are AI ready and that they are comfortable with them. You know, I think we've seen certainly a lot of evolution since we started out with an interest group back in 23 after, you know, after the advent of chat GPT and everyone started talking about it. We had a lot of people that were like on the one end of the spectrum or the other. No, let's hold off. Let's not do anything. And then on the other end of the spectrum saying, let's go fast and let's get going. And we've tried to bring everyone more to the middle and understand, you know, the benefits of the tools and how we can roll them out **Deborah Stine** [33:28](https://youtu.be/3RFNVusazHQ?t=2008) in a responsible fashion. Okay, great. And Vinita, what strategies have you used for engaging workers in AI adoption such as consultation, training, feedback loops, shared governance? **Vanitha Zacharias** [33:43](https://youtu.be/3RFNVusazHQ?t=2023) For AI adoption, we always insist this message that AI cannot happen in vacuum. It takes a village to implement it responsibly. So let's talk about shared governance. And in Ohio, we established this governance model back in 2024. It has like a pyramid with multiple layers. And all the layers are actually responsibility of our state agencies. So they have a lot of AI governance responsibilities in terms of data approval and security governance and agency executive leadership approvals. And the communication between agencies and AI council happened through chief legal counsel. So multiple collaborators and co-collaboration happens through this process. And within the AI council, Katrina talked about it that we established a few years ago. And here like we have our DAS director from representation from governance office from agencies and Katrina and I are part of it. And also like our chief legal data security officers and procurement officers. So everybody is represented and then we make a collective decision on use case alignment with responsible AI principles and ethics. So people may wonder like why shared governance is very important for AI adoption. It's because there's some level of confidence in people's mind. Hey, right people are looking at it. So whoever has responsibility for security, they are looking at it. Whoever is responsible has data ownership, they are looking at it and approving it. So that brings a lot of confidence and this emphasized to collaboration truly helps with adoption and also agency like how we have AI council at the center. Agencies also have established their own governing committees. [35:39](https://youtu.be/3RFNVusazHQ?t=2139) They have data and AI governance committees within their organization who looks at the use case approvals prior to being sent over to AI council. And I think that has helped a lot of adoption too. And also we hear like state agencies have identifying like AI champions or change catalysts within their own organization. And this brings a lot of awareness and engagement and collaboration within those agencies as well and understanding of what processes we are using in Ohio. And so we have also established community type of learning, social learning through our community of practices. We have not only AI community of practice, we have different groups like for data community of practices, for developers and architecture. So everybody talks about AI and shares our best practices. And the best way of learning, especially when you have fear and resistance is through our peers, not through our managers, but through our peers. [36:42](https://youtu.be/3RFNVusazHQ?t=2202) So they share like stories and then we have like sandboxes also established for experiential learning. So whoever has a little bit of inhibition, they can get licenses in sandbox and they can explore and they can learn things at their own pace. And Ohio also has adopted a multi-modal strategy. So we make different models available for our workforce, which means like model that's very aligned with the finance processes, not necessarily for procurement processes or for legal analysis. So different business functions can try out different models and then they can explore for themselves and get more comfortable with AI. And you asked about feedback in your question. So we have established a statewide email called a state of Ohio AI learning email. And we try to get feedback from our workforce on various things that are happening in the state. So example, like they'll say like, hey, we would like to learn more about this particular advanced technique, coding technique maybe, or maybe about how to identify fraud in our business as part of our business processes. And so we take that and then it'll help us to curate the future trainings and even talk to you as what workshops you guys have. So we constantly try to get feedback and then try to evolve our governance processes through that **Deborah Stine** [38:11](https://youtu.be/3RFNVusazHQ?t=2291) those feedback mechanisms. Great, thanks. So we've had so many questions from our audience come in. I'm going to stop here. Hope we can get to some more questions at the end, but I really want to get to the slew of questions that we have from our audience here. And we're about 250, we're up to 250 folks for a while. So as you can imagine, there are a lot of questions. So let's start. A number of questions talked about protecting information, personal identifiable information. How do you, when you're using AI? So let's just start out with that. So Rob, let's start with you, with your kind of thoughts on that, and then we'll move to **Rob Asaro Angelo** [38:57](https://youtu.be/3RFNVusazHQ?t=2337) Nathan, Katrina. Yeah, I mean, this is by far one of the most common concerns I hear, not just from folks I work with, but just in general in the public. And I always try to explain that at least in the government sense, that AI is unlike, I mean, is like other technologies as far as this subject matter. And when I say that, I mean that anything in the government context, especially in a department like ours, which is handling millions and millions of pieces of PII through wage records, health records for temporary disability and family leave and workers comp, PII safety and security is our top priority at all times in every single program. But I think there's some confusion. It's not like we're uploading a bunch of stuff to Claude headquarters, wherever that may be. This is all within our own systems. And the AI tools that we use are based on our own learnings to address some of the other concerns in there about plagiarism and whatever else. I'm not making any judgment calls on there, but I'm just saying that, for example, that appeals tribunal specialist tool, ADA, I talked about, that is trained only on past decisions, judicial precedent, statute, and that kind of stuff. So all the ways we've used AI, it doesn't involve PII. And even if it did, it would already have to be in a secure environment because breaking news, every piece of PII that the government holds, whether it be state or federal, is really in the hands of a vendor somewhere, quite frankly, that goes through the whatever it's called, FedRAMP and that kind of stuff. So whether it be your UI claim system, whether it be your workforce training system, where every workforce board has to upload stuff to the USDOL, that is all hosted by some private vendor somewhere, for the most part, who has to go through these security screens that you're like anybody else does. So I would just say, I don't see it any different than any other technology we're using, that your state, and I am not a technologist, I am not a cybersecurity expert, but I know that so many of our projects, I don't want to say it's a blocker because it's super important, [41:01](https://youtu.be/3RFNVusazHQ?t=2461) but that is what usually takes the most time, is to making sure that all of our PII is protected, not just PII, stuff about our processes and other things like that, are protected. So I don't see AI as being any different than past things we have to deal with with PII. **Katrina Flory** [41:16](https://youtu.be/3RFNVusazHQ?t=2476) I would agree with Rob, you know, we've got those security privacy and data privacy terms in all of our contracts, you know, we're making sure that these tools are in our tenant, the state's tenant, because we don't want, you know, data leakage or anything like that. **Deborah Stine** [41:31](https://youtu.be/3RFNVusazHQ?t=2491) Yeah. Yeah. So this is another sort of question. So this individual talks about that over 150,000 employees have been impacted by AI driven layoffs in 2026, 100,000 in 2025. Why should we honestly **Katrina Flory** [41:49](https://youtu.be/3RFNVusazHQ?t=2509) believe that it'll be any different in government? I mean, I personally see, you know, in Ohio, in Ohio, you know, and Rob, I think you alluded to this at the beginning of your presentation as well, we don't have tons of resources. So I'm not looking to replace these resources with AI. I'm looking to provide our folks with tools so that they can, you know, people that are overloaded, they've got, you've got caseworkers with tons of, you know, cases on their hands, whatever they're doing, if that's in health and human services, is that, you know, insurance, whatever, you know, I don't know people that are just, you know, sitting around with nothing to do. So I just see this as helping them do their jobs, and supporting them, because as we see, you know, and we've all heard about the silver tsunami, I've got some silver myself, you know, if that, if that comes, you know, you know, those positions were also, you know, hiring is not the easiest thing to do either. Here in central Ohio, we've got a lot of competition with, you know, nationwide Ohio State University, Chase Bank, you know, pick somebody, but that's also challenging. **Vanitha Zacharias** [42:56](https://youtu.be/3RFNVusazHQ?t=2576) And we also, I want to add that domain knowledge, institutional knowledge is key for us, like, you know, depending on like, like Katrina said, caseworkers, or eligibility workers, or even like procurement, legal, finance, these are all institutional knowledge that we want to retain. So we are honestly looking to how to upskill them with AI, instead of replacing them with AI, that's not at all where we are coming from. And that is something that we want to also constantly share in our message to build more trust within our institutions. **Rob Asaro Angelo** [43:31](https://youtu.be/3RFNVusazHQ?t=2611) Real quick on that, Deborah, you know, I'm in no way disputing that number of 150,000. And there's certainly been layoffs because of AI, and certainly, obviously, in tech industry, more so than others. One of the things we did in New Jersey, I think back now we're at 26, I think probably in 24, at some point, we sort of want to have a baseline about this question about AI layoffs, or layoffs caused by AI, because there wasn't really great data out there. And quite frankly, can't really trust companies, they don't always want to say why a certain layoff is happening. But just for us in New Jersey, so first, we started having a, when someone filed for unemployment, they get a survey from us afterwards, we added a question in there, was your layout, was your separation, the result of AI? Now, this is self-reported, doesn't know if that's a actual cause or not. Obviously, folks who are answering an online survey are a little skewed as far as reaching out back. But we want to have a baseline. And from that time in 24, through my time leaving in January this year, that number hovered between 10 and 11%. And the only time we saw any kind of noticeable spike, quite frankly, was when there was an AI layoff in the news, even though it wasn't based in New Jersey. So I'm not, just to be clear, layoffs are real. But I do think there's still a correction period going on after all the hiring during COVID. And so I just want to put out there, but I think we should be asking states to try and collect this data. And so I know some states, I think including New Jersey now, have asked on war notices for the companies to say, was the layoff related to AI? [45:00](https://youtu.be/3RFNVusazHQ?t=2700) Again, that's imperfect data for sure, but I think it's important to be measuring that data. So you could have, whether it be individualized rapid response or whatever else, maybe, but I think it's important to have that knowledge. And I don't know that we have a great data set yet about the true cause of layoffs in all cases. And I guess my point is, I don't always believe what the employers say. I think it's become sort of a crutch for some employers to say, oh, these layoffs are because of AI. But I think time will tell. **Deborah Stine** [45:27](https://youtu.be/3RFNVusazHQ?t=2727) Okay, Rob. So somebody would like to hear more about your procurement framework. **Rob Asaro Angelo** [45:32](https://youtu.be/3RFNVusazHQ?t=2732) I saw that question and I got nervous. I am not a procurement expert, but certainly I'm aware of what goes on in procurement. I would just add this. I think, and again, how this isn't just about AI, this is about better technology in government. And I've been a big proponent of agile procurement, agile development, all of your technology projects. We ran the first agile development project in New Jersey with our UI modernization, or second, I should say, Treasury had one. But my point is, during that process, it was hard to go through procurement because there wasn't a, you don't have a, you know, the statement of work is different, the deliverables are different. They're changing at the whole point of having agile development. So not just procurement, but also budget, finance and accounting, people who are overseeing you, auditors, they're so used to looking at something that says, okay, we're going to spend $10 million, we're going to get X, Y, Z. We don't necessarily know what we're going to get when we're doing agile procurement, agile modernizations. The point is, we're going to learn and test along the way. So I guess that same kind of methodology should be applied to AI or any other technological tool. [46:34](https://youtu.be/3RFNVusazHQ?t=2794) But I think just AI is just moving a little bit faster than other things. So we need to make sure that we don't have some new tool and then it takes a year and a half for procurement or whatever function, budgeting, whatever it may be. And then by the time it's finished, that tool is no longer, it's now obsolete. **Deborah Stine** [46:52](https://youtu.be/3RFNVusazHQ?t=2812) So we have several questions that focus more on sort of the ethical aspects. In other words, is the platform they're concerned that AI was built on with unethical practices and then thinking about its ethical use and technology, like how you manage that within the public workspace. So any thoughts on that? **Vanitha Zacharias** [47:20](https://youtu.be/3RFNVusazHQ?t=2840) I think that is something that in State of Ohio's AI Council, we look at the use cases from, definitely from ethical angle, like how it would impact our public workforce and how public users and then how it would impact our employees. So we look at it from those lens, like some use cases, like for example, in corrections or in tax. So there are certain questions that come up from ethical angle, which we work with the agencies for further clarification. Hey, how is this output from AI is being used and who's making those decisions or is human in the loop to verify the information before actions are being taken, especially if it is like sensitive type of use cases. So we definitely look at it from those angles. Yeah. **Deborah Stine** [48:14](https://youtu.be/3RFNVusazHQ?t=2894) Okay. And Rob, somebody would like more information on how AI will create more jobs in government. **Rob Asaro Angelo** [48:22](https://youtu.be/3RFNVusazHQ?t=2902) Yeah. So I think a couple of things. One, I think that as government becomes more efficient and more responsive to the folks that we're serving, there'll be more love for government, which there isn't a ton of right now. That's why I strongly believe that whatever tool we can have in government to make us be more responsive, to share our data better across agencies, to help customers get their unemployment quicker, to help them get permits faster, all that stuff that we hear about with abundance and everything else. All these things have to happen with workers. And I do feel strongly that there's such a default position by elected officials, quite frankly, of both parties, that they want to shrink government. I think that if we can show that government can be responsive and use AI correctly to make it more effective and bring more value to our customers, they'll be more likely to be more pushed for more of that. We love this good service. We want more of it. **Deborah Stine** [49:19](https://youtu.be/3RFNVusazHQ?t=2959) Okay. Great. And Vinita, I expect this question might be for you. I'm interested in thoughts on training that goes beyond basic technology use and instead builds skills that reduce dependence on vendor-provided AI tools. For example, training on how to implement and run local LLMs using a curated knowledge base so we can maintain full responsibility and accountability for AI in our workflows. **Vanitha Zacharias** [49:47](https://youtu.be/3RFNVusazHQ?t=2987) So currently we are in the process of establishing an AI RFP for our state, and we are establishing like a pre-qualified vendor pool. So that's like a life cycle of different phases of AI solution. So as a vendor finishes the implementation, we have established as part of contract knowledge transfer skills. So we are also producing or providing a lot of advanced skills through vendor partnership for our workforce. But as and when a solution is being developed and implemented, make it more reinforcement through your process. Make it like a knowledge transfer of information of those skills from the vendor to your workforce. That way you can learn how to maintain that model, have environments where you can monitor those models and see if there is a data bias that's happening. [50:50](https://youtu.be/3RFNVusazHQ?t=3050) I think slowly we can ramp up our skills, your technologists, and see how to transfer some of your existing skills from software engineers to MLOps engineers or like data engineers into, you know, like two or three different roles to build your AI talent. And so that is the way to build it as and when your organization is building more and more solutions, you start maintaining it instead of just giving those managed services and operational services to your vendor community. That is something that we have to be mindful in order to be upskilling ourselves from workforce perspective. Beyond training, I think doing hands-on is when how we will start building those skills. **Deborah Stine** [51:38](https://youtu.be/3RFNVusazHQ?t=3098) OK, Katrina, I think this question might be a good one for you. Which is, how can we address the differing levels of accountability and responsibility that arise from unequal access to AI tools, both within a single organization and between organizations? **Katrina Flory** [51:57](https://youtu.be/3RFNVusazHQ?t=3117) That's a good question, because certainly, you know, we have agencies, there are have and have not agencies. We all know that based on your revenue source and things like that. And that was something that we were trying to, you know, democratize those tools, if you will, with our training and sandbox environments and making sure that we gave guidance. You know, there was a lot of discussion when when CoPilot first rolled out, you know, how do agencies, how can we help folks make those decisions? You know, there was a thought of, well, just give it to everybody. Well, does everybody actually need it? But how do you make that decision? So in some cases, we encourage folks to roll out kind of like a proof of concept, you know, you know, select some folks and then get feedback from them. And a lot of people did that. So they they picked various areas. They evaluated it, let those teams evaluate it. Did it provide the benefit that they were looking for, you know, based on varying criteria to see if they wanted to roll it out further? So that's a great question and something certainly that we have to grapple with. **Deborah Stine** [52:59](https://youtu.be/3RFNVusazHQ?t=3179) OK, Robin, added the we're since we're running out of time, I'm going to give this last question to you, because I kind of feel it's something you might be able to answer, which is shouldn't all AICOPs working with work with employee unions? What do the unions say about replacing government work with AI and how to define it? If it's something that should be done by tool versus a human. The unions are the only ones who can look out for humans, whereas executives are only looking out for billionaires. **Rob Asaro Angelo** [53:28](https://youtu.be/3RFNVusazHQ?t=3208) I love this question, and I'm going to I'm sort of going to combine it a little bit with a question about AI replacing skills and harming new employees. And I also heard recently about, you know, AI can't come on board and be as effective as a six, as somebody who's been there for 15 years, whatever it may be. So that and the question you start with, Debra, I think are all just management questions and they're going to differ by state. Right. And just like this, like any other tool, if it's in your collective bargaining agreement on your state law that you need to talk to the union about any kind of change in process, then you're going to do that in this process as well. So I think that I think that should be inherent. I personally think it should be inherent in everything you do. I'm a proud union kid, work for AFSCME, which is Katrina talked a little bit about before for her staff. So everything I do is in beautiful worker voice. But I do think it's important, again, that this isn't just like we're going to put AI in some agency and then like, OK, go to work, AI. It's got to be managed like anything else or anybody else, quite frankly. [54:29](https://youtu.be/3RFNVusazHQ?t=3269) If it's a tool that's used, it's going to make sure we're going to make sure that it's used properly in your division with a task that is doing and that you're bringing your folks up to speed on it. And just like anything else, whether you're changing your rules about driving reimbursements or time in the office or anything else, if this is changing your work product or your work processes and that's covering your collective bargaining agreement, then, of course, the unions are going to be involved because that's the process. And if anybody's not following it, they're not following it. That's a whole nother issue, regardless of AI. So I just want to say that AI and the people who are using AI need to be managed like any other workers using any other tool. **Deborah Stine** [55:08](https://youtu.be/3RFNVusazHQ?t=3308) OK, so I'm just going to give you one sentence. We only got like a few minutes left and I still have a few things to do after this. One sentence in terms of what is the single most important takeaway that you'd like our audience to know. Katrina. **Katrina Flory** [55:23](https://youtu.be/3RFNVusazHQ?t=3323) One sentence, you know, I hope that everyone has seen based on what we've talked about, you know, between Benita and I and Rob that all the states, I think, are trying to take a measured approach, trying to be transparent with their workforce and their customers about their use of AI and their vendor partners. It's certainly a powerful tool and we want to use it for good, but it's a good work for Ohioans and for our teams. **Deborah Stine** [55:49](https://youtu.be/3RFNVusazHQ?t=3349) OK, and Benita. **Vanitha Zacharias** [55:51](https://youtu.be/3RFNVusazHQ?t=3351) Establish an organizational culture that promotes continuous learning. **Deborah Stine** [55:57](https://youtu.be/3RFNVusazHQ?t=3357) OK, thanks, short and sweet. And Rob. **Rob Asaro Angelo** [56:00](https://youtu.be/3RFNVusazHQ?t=3360) Keep asking these great questions and addressing the elephants in the room, because if you don't ask the questions, they're not going to be addressed in the right way. So whatever level you're at, make sure you're talking to your boss, the folks you're the boss of and your colleagues about all these things. **Deborah Stine** [56:15](https://youtu.be/3RFNVusazHQ?t=3375) Yes, well, thanks so much. And speaking of our future sessions, the one that we have coming up next, next week, is about power, say, power, place and AI, energy, environment, AI data centers. I saw there was a question about that. So, hey, stay tuned for next week. That session, all of these will be recorded. I'd just like to thank our speakers. Please give them a virtual round of applause. Katrina, Benita and Rob for doing a great job in answering these many challenging questions. And there's like more to come. The Christine is going to put information on the workshops that I mentioned into chat besides the one on energy. Again, we have one that's on people who have disabilities, accessibility issues, which is also a big issue. Please make sure you fill out the post-workshop survey and suggest topics you'd like us to cover in the future. That's how this session came about. And that is how we would like to generate future ideas for our workshops. [57:15](https://youtu.be/3RFNVusazHQ?t=3435) And I will just ask, somebody asked to look at the poll results. I'm hoping Christine can maybe pull those poll results up one more time while we're saying goodbye. So thanks, everybody, so much for attending. It was just great discussion, great and lively issues and more to come next week. Thanks, everybody. **Vanitha Zacharias** [57:38](https://youtu.be/3RFNVusazHQ?t=3458) Thank you. * * * Original page: https://innovate-us.org/beyond-displacement-building-an-ai-ready-workforce · Licence: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/), as stated on innovate-us.org