What Artificial Intelligence Means for State, Local, Tribal & Territorial Gov't
David Orcutt: Then we can. David Orcutt: We can get started. David Orcutt: So hey I want to welcome everyone to today's panel discussion on artificial intelligence, so my name is David orkut and i'll be acting as the technical host so.
David Orcutt: Just as a reminder, please, please remember to post any questions you have in the Q amp a pod so that's at the bottom of your screen you just click on the Q amp a pod and you can type in your question. David Orcutt: we'll get that to them at the end of the discussion so I want to welcome today's moderator David trey. David Orcutt: trey Brown, who is the director of the innovation division within the office of information technology category ITC with the general services administration's federal acquisition service. David Orcutt: As an organization within fs ITC provides access to a wide range of commercial and custom it, products, services and solutions.
David Orcutt: Mr taban join GSA in 2013 as a contracting officer he managed directed and coordinated procurement activities for the pre award and post award actions within GSA strategic solutions division. David Orcutt: He is currently serving as the director of it sees innovation division supporting the federal acquisition workforce. David Orcutt: federal agency partners federal supplier Community and also presidential and omb initiatives all to make innovation. David Orcutt: And emerging technology services and solutions available on GSA is acquisition vehicles, David has over 30 years of contracting and acquisition experience, specifically with it hardware, software and professional services so with that i'm going to turn it over to you, David. David Taybron: Thanks David Hello everyone and welcome today to today's sessions on what artificial intelligence means for state local tribal and territorial governments. David Taybron: i'm pleased to have a panelists have experts for today's presentation, I am David tabor on your host and I would like to introduce our three panelists Sam Navarro Eric Ealing and Sabrina Mohammed.
David Taybron: Sam to Martin Sam Navarro sorry Sam is the senior procurement advisor senior program advisor within the office of technology category. David Taybron: In GSA is federal acquisition service, Mr Navarro, is responsible for planning and implementing government i'm sorry customer development strategies. David Taybron: For ITC including government wide acquisition contracts and business lines in the ITC portfolio previously Mr Navarro served as the as the director of the commercial strategic solutions division. David Taybron: Joining GSA in the fall of 2014 Sam has held several positions in the Agency. David Taybron: As the wireless mobility program director. David Taybron: Contracts modification manager and contracting officers representative for the third largest network and telecommunications contract in the Agency he's also held the position of federal agency manager assistant agencies modernizing their network infrastructure across the United States.
David Taybron: Our next panelist Eric Ewing has over eight years of experience in data analytics and artificial intelligence initiatives. David Taybron: Eric has served as a senior advisor to multiple chief data officers i'm sorry multiple federal chief data officers, providing leadership guidance and expertise in the enterprise modernization into enterprise modernization initiatives within this Center of excellence. David Taybron: Mr Ewing has consolidates and organized long and short term strategy for data and Ai while leading initiatives for the management of data as a strategic asset. David Taybron: Prior to joining the Center of excellence Eric was a data scientist and a senior consultant leading key analytics programs Ai research and development, and it supply chain security initiatives both private both with private and public sector partners. David Taybron: Our last panelist here is Sabrina Mohammed she's she is the President says i'm sorry she's a presidential innovative innovation fellow. David Taybron: detail to GSA is technology transformation services she's led the artificial intelligence community of practice, which brings together federal employees to support the practical implementation of responsible Ai within the Federal Government.
David Taybron: Prior to being a president's are innovative innovation fellow Sabrina worked at Microsoft she's recently worked on word. David Taybron: Building Ai powered ways for information workers to collaborate and staying on top of personal productivity sabrina's mission as a product leader is to incorporate ethics and human centered design into technological conversations. David Taybron: So with That being said, with our introductions let's kick off today's discussion. David Taybron: So what is.
David Taybron: artificial intelligence artificial intelligence is commonly referred to as Ai. David Taybron: It has been in the news frequently and can often be misunderstood. David Taybron: To our panelists, can you please help us understand what Ai is what are your thoughts Sabrina. Sabrina Mohamed: Thanks David i'm so to take a first stab at that question, I think, a good like to the question of what is Ai you can also ask what isn't Ai um since Ai can fit into. Sabrina Mohamed: pretty much every single sector use case of technology or not even if it's a technology of society and a definition is artificial intelligence is referring to all of the computational techniques that can stimulate human cognitive abilities.
Sabrina Mohamed: And so, Ai in the future will transform most, if not every aspect of society humanity people's behavior and. Sabrina Mohamed: A good way of explaining it is, I mean there's so many different use cases like Ai is just a way of making sense of data. Sabrina Mohamed: For to help human beings, so we talk about automation we talked about different virtual chat bots that are interacting with people, those chat bots that's an example of a use of Ai technologies. Sabrina Mohamed: If we're talking about certain types of robots if we're talking about gathering insights from large amounts of data, you can use Ai in order to do that, and I think the term is so broad and encompasses so much that.
Sabrina Mohamed: That there's no one clear example that I can give it's an Ai is often it's not it's not these moonshots it's not Ai is a cyborg Ai is. Sabrina Mohamed: Like a computer taking over the world, Ai is IBM Watson it's more like Ai can plug into literally anything I mean if you're talking about if you have a smartwatch and apple watch that uses artificial intelligence. Sabrina Mohamed: If you're using a computer if you're interfacing with a financial institution such as a bank, the way that they. Sabrina Mohamed: Sometimes will come up and say hey, we think that you might have a fraudulent charge because something is outside of your normal patterns of. Sabrina Mohamed: Purchasing that's an example of using it, so it really does if you can think of a use case a business problem there are ways for Ai to plug into it whatever they thought, maybe. David Taybron: Very good, thank you for that um how about you Eric Could you give a you give us your perspective on what is Ai and help us understand it.
Eric Ewing: yeah absolutely i'm. Eric Ewing: Sabrina is dead on Ai is overly broad sort of topic, but I think the simplest way to describe it, and they bash me in academia, if I said this, but is the ability for computers to be able to teach themselves given given human parameters so ultimately. Eric Ewing: Ultimately, to go a little bit further, I want to sabrina's examples there. Eric Ewing: If you're if you get a game on your on your phone or whatever it says your credit card had an. Eric Ewing: may have had a fraudulent charge it's because the computer and learned about your relative patterns.
Eric Ewing: and learn that something was anomalous outside of that so you know you bought a jet ski which congratulations you bought a jet ski but you don't do that every day, so you know they're gonna they're gonna Ping it they're gonna you know give you a paying for that. Eric Ewing: So there's a lot of examples of how that comes up outside of personal finance and government as well and i'm really excited to get into that but, at the core it's The ability for computers to learn to to teach themselves. David Taybron: To me, thank you for that um Sam how about you. Sam Navarro: So i'll take a market perspective, look at what Ai is and it's a very general concept but. Sam Navarro: From a market perspective, I think, Ai is the ability to replace or enhance human capital resources with computer resources that can mimic or enhance human. Sam Navarro: interactivity you know I think we've all been there and i've worked with a lot of agencies and state and local entities that have tasks to do.
Sam Navarro: just not enough bodies to do the work right and so Ai is an opportunity where you can leverage a tool to enhance the ability of people that you have to do the work. Sam Navarro: Or, as we as i've heard said, sometimes take the robot out of the activity right. Sam Navarro: Some some single domain tasks that are probably Wednesday and kind of standard repeatable can be enhanced down or done by computer. Sam Navarro: Freeing up humans to do more, you know intellectual work that's multi domain in purpose so it's a great opportunity to gain efficiencies sometimes depending on what the solution is right. Sam Navarro: Other times, it can be just an opportunity to enhance a human's ability to make a decision by a computer doing calculations are coming to conclusions. Sam Navarro: That would take a human quite some time to do you know, I was just reading an article by the Mayo clinic where they deploy an Ai solution to just look over mris done by by doctors to look at cancer patients or potential cancer patients and percentage of positive.
Sam Navarro: Findings within that space with not only was increased but it wasn't hands, because the Ai could capture things that doctors potentially could overlook. Sam Navarro: And then the survivability rate from patients increase because they were able to capture the cancer at a much earlier stage than a later stage if it was overlooked. Sam Navarro: By patient and all the data that goes into Mrs i didn't even know it was over 100,000 images go instead Mr a I so having a computer or an Ai solution look over all that data all that content. Sam Navarro: Definitely increases efficiencies because it could do it a lot faster than the human, I can but also as Eric alluded to the ability to learn. Sam Navarro: As it finds different things on how those things can look like is really limited as well, so from a market perspective there's a lot of opportunity within the public sector. Sam Navarro: To really leverage tools and as products and solutions to do what we do for citizens on a daily basis, but do it more efficiently and effectively.
David Taybron: Very good, thank you, Sam. David Taybron: i've heard it said that data is the fuel for Ai so applying the software programs and the tools that we have the the artificial intelligence tools. David Taybron: And you guys spoke very eloquently about how we can use these tools to to manage and to draw conclusions and to learn. David Taybron: From those data sources so Okay, thank you for those responses, the next question that we have is recently the national state of CIO National Association of state cios were asked.
David Taybron: Which emerging technologies, they believe would be most impactful within the next three to five years and 65% of those polled named Ai artificial intelligence, could you talk about what you see as the future for Ai and Sam let's start with you, please. Sam Navarro: yeah thanks for that question, and I think you're spot on Dave the lifeblood of Ai is data and we've seen a lot of great use cases within the public sector kind of enhance the ability of us to serve citizens today and partially around that is and public safety. Sam Navarro: Also in Smart Cities smart building concepts The ability for an algorithm or computer to take in a lot of data and provide that to first responders quickly. Sam Navarro: has been part of the use cases that we've seen deployed within the public space and so now that we're getting faster networks like five g and access to a lot more data it enhances that ability to.
Sam Navarro: To really serve the public in a very special way so we've seen things happen in the past, such as utilities having challenges, such as in flint Michigan where. Sam Navarro: We had water that wasn't safe to drink well think about the ability of having sensors out there that continuously test utilities to make sure the water is safe to drink and then authorities ability to quickly respond to. Sam Navarro: outcomes that may not be positive for the for the public that's just kind of one example, one use case where I think the outcome that the cios came to was right there's a lot of value to be driven by. David Taybron: You saying some work life balance that's all right, thank you, thank you for that same Sabrina how about you could you could you give us how you see your future well as future from your perspective. Sabrina Mohamed: um yeah, thank you for the question.
Sabrina Mohamed: um so what comes to mind is there is this technology hype curve by gartner and whenever there is a new emerging technology, you know it starts out at the beginning of the curve, where you have an innovation. Sabrina Mohamed: That innovation trigger and then people have immense expectations of it and then. Sabrina Mohamed: kind of going down a little bit you have this disillusionment, then you have enlightenment and then you kind of reach a plateau of you know, productivity, that is a fully integrated technology moving forward. Sabrina Mohamed: I think, where we are in that curve right now is we've had the innovation trigger we're kind of at the top of the peak of inflated expectations with Ai and we're just figuring it out. Sabrina Mohamed: And soon enough we'll get toward like setting up and establishing organizations in a way, so that we can actually use that technology, but where I see the near term future is. Sabrina Mohamed: Organizations Federal Government state local territorial tribal governments figuring out how to structure their organizations to.
Sabrina Mohamed: embrace Ai to learn more about what data exists within that organization to figure out how to integrate it into existing business practices so that it's not just like a bolt on but rather it's actually part of the business. Sabrina Mohamed: And how it operates and how it thinks about use cases and really just like setting up for success thinking about the people who are serving how to serve them with Ai. Sabrina Mohamed: I think right now or in the near term future in this phase of figuring out how to organize and manage data so that we can. Sabrina Mohamed: Supply Ai solutions into it and a lot of that what needs to happen, for that is just developing the Ai workforce. Sabrina Mohamed: And what I see in more of the future further term is solving more business challenges with Ai and having it be something that's like fully integrated and fully normalized into how we think about problems.
Sabrina Mohamed: In the same way that. Sabrina Mohamed: they're just like standard ways of thinking about problems standard business processes standard workflows Ai is going to be a part of that in the future. David Taybron: Very good, thank you Sabrina Eric we'd like to hear your perspective on the future for Ai. Eric Ewing: yeah in my role I work with a lot of different federal departments and agencies on Ai implementation and ultimately. Eric Ewing: There is, you know the grand vision of Ai in X number of years you don't want to put a time frame on it, but. Eric Ewing: Where automated systems are much more regular in the government space and they currently are.
Eric Ewing: But in the near term I think there's a lot of low hanging fruit, a lot of a lot of real opportunity for Ai to make a major impact and what we do. Eric Ewing: And to david's point earlier about data being the lifeblood of Ai ultimately um the areas in which that low hanging fruit exists, are ones, in which we have data like data rich structured data that is like extraordinarily rich and has a long history, within our organization so. Eric Ewing: I think government financials or one particular place where I think Ai is a is right for deployment and there are a lot of different use cases in terms of de risking your financial portfolio. Eric Ewing: To using very simple Ai to project budget to predict to predict any financial changes that are happening. Eric Ewing: And you know this has been done on Wall Street, for a long time we as government organizations have a long history of collecting these these this budget information.
Eric Ewing: Now it's not always as clean as we need it to be so, you might have to clean it up, etc, etc, but it is in terms of in terms of data sets that are being collected, they are very, very structured. Eric Ewing: likely, the most you'll find additionally in terms of grants management, I think there's a lot that can be done there. Eric Ewing: Simply because grants management is often outside of the financial aspect, even the applications, the reviews they're often very structured very structured as well, and these are things that state local federal governments are constantly putting out as grants. Eric Ewing: And I think.
Eric Ewing: With machines ability to be able to process. Eric Ewing: Language getting stronger and stronger every day, although if you use siri there's still a long way to go. Eric Ewing: But you know, there is a natural language processing this this specific Ai technology, under which you know it could.
Eric Ewing: computers are able to to read documents, and that is getting stronger every day, and I think the ability for for Ai to have a major impact and streamlining our grants management processes. Eric Ewing: is going to be extraordinarily important going forward I think it's something that we that we as. Eric Ewing: As government needs to be working on right now, I think there are opportunities in the in the immediate future. Eric Ewing: To take advantage of Ai to to help us streamline that process, but I think the long term impact is going to be going to be incredible, but I think the real short term impact will be sort of non trivial as well. Eric Ewing: Lastly, I have to touch on it, because it was brought up and i'm a Detroit guy i'm going to try.
Eric Ewing: Like the ability for. Eric Ewing: The ability for Ai to improve the way we distribute public goods generally. Eric Ewing: And in a more efficient way, whether it be water or housing. Eric Ewing: Industry sees this. Eric Ewing: I know all the large mortgage firms, the way homes are underwritten the way homes are appraised are often based on historical data that leverages artificial intelligence or machine learning.
Eric Ewing: But there are gaps in that historical data that often entrench biases that have existed for a really long time, so one of the things that. Eric Ewing: One of the things in the provisioning of public services, and then the provisioning of. Eric Ewing: Ultimately, shared experiences that most people go through throughout their lives are vast majority of people in the country go through throughout their lives there's also opportunities to use Ai as a forcing function for equity.
Eric Ewing: and other things of that nature, and I think we, I think we have to keep that at the forefront of our minds as well, thanks for bringing up floats and I appreciate that. David Taybron: Thank you, thank you for that Eric i'm switching gears just a little bit to something that's very that's extremely an. David Taybron: ancillary but aligns very closely with Ai is cyber security it's cybersecurity is on everyone's mind these days as it should be rightfully so and they I many times, and most of the time is cloud based. David Taybron: For the most part um in terms of our GSA solutions are the GSA vendors Hilton some standard when it comes to security and Ai in cloud solution services i'm sorry cloud services. Sam Navarro: So that's a great question and the reality is is the solution has to reside on some. Sam Navarro: Some certain level of infrastructure right whether it's on the cloud on the edge or on a device, and the answer to that is yes, there are cyber security standards.
Sam Navarro: embedded in GSA contracts and the one that's very familiar to a lot of folks out there is fed ramp. Sam Navarro: To make sure that cloud providers that are providing solutions off of GSA contracts reach a minimal standard of security also for transport, you know we have. Sam Navarro: fitness standards from high to medium to low, to make sure that carriers that may be providing access to that solution also secure the network, so we have cyber security embedded across our solutions from the device from end to end. Sam Navarro: solution for that reason, because we understand that the data, whether it's in transit or whether it's a rest or being a computer needs to be secured and protected.
David Taybron: Thank you, Sam. David Taybron: We see you're raising the next generation of data scientists there, so we appreciate that as well, so. David Taybron: So Eric do you have any do you have. David Taybron: Any any any feedback on that as well. Eric Ewing: yeah I think ultimately.
Eric Ewing: When looking at emerging technology we're looking at Ai, I think, for a long time we've talked about it but technology space like having cyber cyber security built in, from the outset. Eric Ewing: And while that's not specifically my area of expertise. Eric Ewing: Being cyber security, I think one of the things that we try to do when evaluating new technologies is make sure we have a cyber expert in the room. Eric Ewing: and ask what steps have been taken along the way to ensure that data data, especially, especially, as you know, it's being put through an automated automated system or through an Ai system is. Eric Ewing: remains as secure as we expect, given the situation now.
Eric Ewing: I think there's a I think there's some great tools out there, including fed ramp. Eric Ewing: That we look to in as the federal government. Eric Ewing: You know, fed ramp authorizations and. Eric Ewing: and tools of that nature but baking cybersecurity in from the start, is always is always a key we as a government have a responsibility to bake it into the system, but if we're buying pieces of the system or for buying the system as a whole.
Eric Ewing: That has to be one of the one of the legs of the stool that we're evaluating in terms of costs, security and productivity so. David Taybron: Okay, thank you Eric you know, having having that cyber security baked in and just making that know paramount in a solutions that are deployed is is is one of those special is one of the primary considerations Sabrina do you have anything to add to that as well, please. Sabrina Mohamed: I know I do not, I think Eric and Sam said it quite well and. David Taybron: Thank you for that. David Taybron: So our next question um. David Taybron: If states in local governments and and our audience here wanting to move towards a universal broadband policy, how would that impact Ai and I think Sam that specifically for you.
Sam Navarro: So I had the privilege of working in the mobility program prior to this position and 5g was a big. Sam Navarro: topic and it's a big topic today right as we try to bridge the digital divide and bring broadband areas that are at a disadvantage today. Sam Navarro: And the value out of the 5g network was that it promise lower latency higher data rates and the ubiquity can unleash endless amounts of use cases right but. Sam Navarro: The interesting piece that I think Erica Erica Sabrina would agree, is a lot of the use cases we talked about a 5g couldn't be enabled without Ai right when you talk about autonomous vehicles. Sam Navarro: And all the other use cases that we can discuss really can't can't be a least unless there's a smart computing capability behind it and so that's the ability and the power behind broadband policies across the United States and we're working with.
Sam Navarro: The City of Vegas we're working with various entities in the State of Texas and North Carolina to provide them the support they need to not only deploy broadband, but then think strategically about what solutions as can unleash in the sector of education. Sam Navarro: As we said, public safety or other verticals within that space as well, so the reality is the network is kind of the backbone, and in order to at least some of these use cases. Sam Navarro: You really need that network to to be able to empower and move the data to as quickly as possible right to have real life real time results. David Taybron: Very good, you know, having that robust backbone take to take advantage of this data to get the fuel to to the Ai solutions back at the the five g and the broadband policy is paramount, so thank you for that same our next question how would you to the panel, how would you ensure equitable. David Taybron: program with Ai How would you ensure an equitable program with incorporating an artificial intelligence solution. David Taybron: Addressing instances of bias as you've mentioned earlier and privacy concerns so Sabrina could you start us out here, please.
Sabrina Mohamed: Sure, and i'm also really glad that we're asking the question about equity and I think that with any conversation about Ai it needs to be in the forefront. Sabrina Mohamed: As part of the group that I lead, which is the federal Ai community of practice, we have a working group specifically focused around responsible Ai and so it's something that we care about a lot within Federal Government, the White House curious about a lot as well there. Sabrina Mohamed: They put out a piece talking about how we need Americans need a bill of rights for an Ai powered world specifically going over what are the different rights that citizens can have guaranteed. Sabrina Mohamed: Before going into some of the things that we can do to make sure that Ai does equitably serve the public I think it's important to think about and mention what happens when it doesn't semin Eric mentioned flint Michigan as an example. Sabrina Mohamed: Of. Sabrina Mohamed: or housing discrimination as examples of.
Sabrina Mohamed: If we're training on the wrong data, then those systems might make decisions that that reinforce existing problems in society so to give an example. Sabrina Mohamed: As I said, a training machines based on earlier examples what that does is it can embed past prejudices and enable any present day discrimination. Sabrina Mohamed: So hiring tools, for instance, these are automated tools that can look through people's resumes and they can reject or move forward resume and so.
Sabrina Mohamed: What that can do if it's learning off of current patterns is they might reject applicants who are dissimilar from existing staff so, for instance, if you have a computer science. Sabrina Mohamed: department or team where there isn't gender parity and there aren't enough women on that team, then that automated system might be rejecting. Sabrina Mohamed: Applicants on the basis of gender, which is gender discrimination. Sabrina Mohamed: Just based on existing patterns and so there's a lot of problems that might emerge around equity with computer science or sorry with them artificial intelligence some ways to combat that. Sabrina Mohamed: One is making sure that teams are diverse, so the people who are.
Sabrina Mohamed: Who are purchasing a is the people who are implementing it playing I making sure that there is representation that is similar to the representation of the American public I think that's one really important tactic. Sabrina Mohamed: Another is asking questions often and repeatedly so a good starting point to responsibly implement Ai is make sure that you're asking questions around key decision points and to ask them early and to ask them off and. Sabrina Mohamed: This is guidance that our community of practice and Center of excellence gives out is making sure that you're asking the same questions over and over and those answers might change as the team learns. Sabrina Mohamed: But you need to make sure that your the task at hand is on the people who are actually deploying the tools to make sure that.
Sabrina Mohamed: It is responsible, and it is trustworthy we hear about responsibility, I and trustworthy i'd one thing that i'd like to emphasize is there's no such thing as a. Sabrina Mohamed: Good day or bad Ai at the end of the day, it's just code that people develop and it can be used responsibly by us, or it could be used irresponsibly if there's not. Sabrina Mohamed: enough attention put into who are the people who are serving and how are we serving and so, in order to ensure that it is equitable for the American public that task is really.
Sabrina Mohamed: on us to consider, to make sure that we are asking questions each and every single step of the way of what data is going inside of this. Sabrina Mohamed: What potential risks might come up. Sabrina Mohamed: And is there is there a human in the loop is there, someone checking up on this system if something does go, not according to plan. Sabrina Mohamed: it's important to make sure that systems, even if they are autonomous, there is an actual human being who's there who's taking accountability for whatever decisions are being made from that. David Taybron: Thank you so much for that Sabrina Eric could you give us your perspective on ensuring equitable programs are baked in using your word again baked into the Ai Ai implementation.
Eric Ewing: Yes, certainly, and I mean i'm going to take much the top of you know, I take much their approach that Sabrina does this, which is ultimately that. Eric Ewing: We have to create a appropriately and attachment system for Ai systems, we have to create appropriate management structures at the end of the day. Eric Ewing: When developing it there's kind of best practices for creating management structures that maintain accountability and main you know, keep operations and maintenance for any system rolling.
Eric Ewing: I think in Ai maintaining that that it best practice for management structure is. Eric Ewing: Perhaps even more important, simply because, when a mistake is made in an Ai algorithm it may not break the system, but you may just be getting the wrong outputs so any decisions you're making on what they algorithm without putting or any decisions at the Ai the Ai system itself is making. Eric Ewing: may not even be detected without the proper management structures over it, so you could you or the or the automated program maybe making decision, maybe, making the correct decisions without even knowing it. Eric Ewing: If you don't have that proper management structure in place and accountability and traceability at every step so.
Eric Ewing: I would ultimately I would ultimately agree with Sabrina on, that is, we have to create, we have to create accountability through through management structures and make sure there's transparency every step of the way as we're as we're in development of Ai systems. Eric Ewing: Good Thank you. David Taybron: and Sabrina, thank you for dropping your your the Ai. David Taybron: The it modernization Center of excellence, your community of practice on your Ai community of practice a link in the chat so please everyone, please take advantage of that and and look what we have.
David Taybron: and Sabrina and Eric have in terms of their community of practice seems to be very robust in and right out here and very timely so so thank you for dropping it there Assam could you give us any considerations, you have for addressing instances of bias and privacy concerns as well. Sam Navarro: um well I tell you what it regards to. Sam Navarro: The equity within Ai I think Eric and Sabrina were very. Sam Navarro: articulate and the way they described it and I probably couldn't add much more, other than to say I heard an analogy, just to for folks to keep in mind. Sam Navarro: He doesn't have any empathy right or it doesn't have compassion, or some of those things that are important in determining whether those are the outcomes that best me. Sam Navarro: Then needs of the citizens we're trying to serve and so really thinking about equity and making sure that it serves us.
Sam Navarro: In the end, is truly where the rubber hits it hits the road just kind of like when we Pave a road right and I heard somebody way more smarter than I am get used this analogy. Sam Navarro: We don't care if there's an ECHO there right it's not that we can't dance but we're paving the road right, and so, in the same vein, we just need to make sure that you know, Ai which doesn't have emotions tied to it just. Sam Navarro: takes into consideration is all the factors we as humans would like it to take into consideration before it's deployed or while it's deployed to serve a use case. David Taybron: There, thank you, Sam.
David Taybron: So let's move on to our next question, do you as a panel, do you think that the State local tribal and territorial governments could benefit from a federal a standard framework. David Taybron: on how to best manage understand and mitigate the risk surrounding Ai much like is much like of what's being worked on at the federal level um how about you, Sam. Sam Navarro: So, again i'll come at this from a market perspective I think there's a lot of use cases that all of us could benefit from in the public sector and thinking about this from a shared service model or concept, I think, would be phenomenal for us as. Sam Navarro: public servants to kind of address some of the things we would like to do TTS is a huge advocate to help not only. Sam Navarro: federal agencies, the state of local entities as well they're looking to. Sam Navarro: deploy this space, but I would say is let's think about the efficiencies and the effect of missing coming together and buying a solution that may work for various state and local entities.
Sam Navarro: Together, or a multitude of staying local entities, where we can consolidate the buying power and use that by power to drive down the price of a lot of these solutions where we're trying to consume. David Taybron: Thank you, Sam how about you Eric. Eric Ewing: yeah i've always looking for broad based broad based frameworks and solutions that to us as.
Eric Ewing: As guideposts ultimately I think one of the things that we see swirling in the internationally I space and then you know definitely within the United States as well, is is what does. Eric Ewing: Audit look like for Ai systems, I, I am not aware of any robust audit mechanisms for Ai systems to ensure all of all of this sort of things we've talked about in terms of cyber security equity. Eric Ewing: fidelity of output fidelity of input.
Eric Ewing: But I think one of the things we're looking forward to looking forward to, as well as, and I know i'm stealing sabrina's. Eric Ewing: comment here, but does the dentist Ai risk management framework to come out and I think that'll be that'll be an invaluable guidepost the the nist the mist documents for cybersecurity are the foundational documents there. Eric Ewing: And for those who are you know technologist, and this is the National Institute of standards and technology within the Department of Commerce and they are the rules making body for for it and. Eric Ewing: Science and Technology writ large, but for it, in particular, in this case so looking forward to looking forward to this Ai risk management framework. Eric Ewing: As well as seeing you know evolution in the audit space, I think I think there's a lot of work that's being done, and a lot of good work that will be done going forward in that space to help us.
Eric Ewing: Hold Ai systems and and our ourselves accountable for managing systems so. David Taybron: Very good, thank you Eric Sabrina could you give us your your your feelings on having a standard API framework for local governments. David Taybron: And Federal Government and all governments as a whole. Sabrina Mohamed: yeah um. Sabrina Mohamed: I similarly Eric you can steal my thunder.
Sabrina Mohamed: Similarly, I am so excited for this, the NIS Ai risk management framework, I think this will be foundational in in paving the path toward regulations and the future for what Ai should be across a variety of different sectors and I know that nest has been working. Sabrina Mohamed: Really thoroughly on this it's like when this creates standards they spend years on, it to make sure that they are really good standards they solicit feedback from industry from academia from. Sabrina Mohamed: federal agencies to make sure that they're really representing the end of the folks who are going to be creating these technologies.
Sabrina Mohamed: And so i'm just i'm super excited for this to come out it covers principles around and transparency, accountability, fairness during all of the different phases of the life cycle, including design deployment. Sabrina Mohamed: usage and testing of those technologies, in addition, as Eric mentioned auditing Ai systems that's going to be really important in the future. Sabrina Mohamed: There are a variety of different algorithmic impact assessments that are out there, and so that work is being built upon currently as we speak, and there's a lot coming out and also a lot to be done a lot of opportunity to also adopt some of these measures at a State level or. Sabrina Mohamed: more of a local level territorial tribal.
Sabrina Mohamed: And one other things that I might mention is the Department of Defense and. Sabrina Mohamed: That they published recommendations on the ethical use of artificial intelligence and so these are other standards to look forward, they define five major principles for upholding legal, ethical and policy commitments and those five are responsible equitable traceable, reliable and governable. Sabrina Mohamed: As another resource.
David Taybron: Very good, very good, thank you for that, in terms of finding. David Taybron: Some of the some of what you've talked about. David Taybron: Is there a way that we can provide will probably follow up David with with some of these some of the resources as well and it probably here as well, but but I don't want that to go unrecognized that the there, there are a lot of resources out there that we can add state and. David Taybron: State and local governments can utilize to to kind of fraud that as a framework for their their standards until the federal standard gets put into place so okay um so. David Taybron: just another question on for for state local tribal and territorial governments have pointed towards incorporating Ai as a top priority, what are some of the existing use cases for incorporating Ai enabled solutions.
David Taybron: For these agencies. David Taybron: And how can these agencies use GSA cooperative purchasing program not not a subtle plug for the GSA right there, but how can the agencies, use the GSA is cooperative purchasing program to help with their procurement needs. David Taybron: Eric if you could you could start us off, please. Eric Ewing: yeah i'm going to steer away from the procurement needs but i'll stick to the tech, which is my domain. Eric Ewing: I think in terms of. Eric Ewing: In terms of looking for use cases in I.
Eric Ewing: First. Eric Ewing: I asked herself a couple of questions. Eric Ewing: Is what's the business problem or where's the business opportunity if there's not one tell us Ai if there's not a problem you don't need to don't need to hit anything with a hammer to make yourself feel better, but if there is, if there is.
Eric Ewing: Definitely. Eric Ewing: Think about where the where the data rich environments are so I talked earlier about financial grants management i'm sure there are others within your organization think of where the data rich environments are where you'll be able to go back 510 years of have quite a bit of data. Eric Ewing: No matter how messy it is and be able to be able to clean it up and create meaningful. Eric Ewing: Create meaningful Ai systems, out of it and, ultimately, like as long as that data exists that you have confidence that that data is is quality has high fidelity will continue to be served up I those are always use cases where you can start thinking about what Ai means.
Eric Ewing: And what I could do for you there to help you either predict what what might happen going forward based on some of the past indicators might help you automate things moving forward for these big complex systems as well, based on what you've seen so one find find a business problem. Eric Ewing: or a business opportunity. Eric Ewing: Make sure you have a data rich environment in which to deploy Ai. Eric Ewing: And then ultimately.
Eric Ewing: One of the things I think we have to talk about is investing module early So there are a lot of things that have to happen to create a quality Ai system, you have to have good clean data. Eric Ewing: you're probably going to have to invest in creating good clean data, you might have to rethink about the way you capture data. Eric Ewing: Data data capture one of the biggest things that Ai is going to change is the way we capture data across all domains of society. Eric Ewing: Simply because we have to make it digital make it as clean as possible because Ai we're going to start using it more and more, and without those things. Eric Ewing: It doesn't really work so find finding those business opportunities data rich environments and places that you want to invest module early you're willing to take it a little bit at a time and and. Eric Ewing: and take your successes, one by one, I think, are the three things I look for the most when i'm dealing with other federal federal departments and agencies and and that's my recommendation for for sort of all organizations looking to go down this path.
David Taybron: Thank you Eric. David Taybron: Sabrina could you give us some some use cases. Sabrina Mohamed: i'm sure yes. David Taybron: or no discussion of it i'm sorry I didn't I didn't mean that limit. David Taybron: limit your. David Taybron: ability dances.
Sabrina Mohamed: Please um so when considering use cases I mean Eric I hit the nail on the head, the only thing that I would add, is. Sabrina Mohamed: validate with users holiday with a real like people who you're serving. Sabrina Mohamed: Actually interview users, for your Ai projects, I think, very often in organizations leadership program managers decision makers, create solutions for problems without necessarily. Sabrina Mohamed: validating that first with the actual application users, and so the best way to incorporate that is through human centered design practices toward. Sabrina Mohamed: by talking to people really early on, before anything is actually built to validate whether this is actually a problem that they encounter in their day to day realize. Sabrina Mohamed: Make sure that it's a real problem and then.
Sabrina Mohamed: And then, once you know for certain ones you have a lot of validation that this is actually a real worthwhile problem use case that you're solving then figure out Okay, how does we solution it what are different ways of going about it. Sabrina Mohamed: To really go from prototyping then to a fully fledged application and to make the best use of taxpayer dollars biggest bang for buck. Sabrina Mohamed: In terms of actual use cases, as I mentioned earlier it's not what is Ai is what is an Ai there's so many different things that can. Sabrina Mohamed: hey. David Taybron: i'm sorry I thought we lost you but. Sorry.
Sabrina Mohamed: I can be used for a variety of different governance use cases so Examples of this include internal management that's internal management of agency resources. Sabrina Mohamed: and maintenance of systems that can be used for public services and engagement so tools to actually better engage with the public or to facilitate communication. Sabrina Mohamed: It can be used for regulatory regulation purposes, for example, the Federal Government has regulations.gov where we solicit feedback from the public for any proposed regulation.
Sabrina Mohamed: Often there are thousands of comments come from that, and so the problem statement, the challenge and the opportunity is how do we understand. Sabrina Mohamed: The sentiment of what's being communicated, how do we get the main points and then communicate back to the subject matter experts. Sabrina Mohamed: And we can do that through artificial intelligence techniques called natural language processing that actually looked forward tax and then derive insights and draw insights from that. Sabrina Mohamed: It also can be used for as we've kind of mentioned before, potentially looking I violations such as fraud and the financial services use case of anything that's outside of establish patterns and norms just to flag it as hey this might be something potentially concerning. David Taybron: Very good, thank you, thank you so much for that and Sam if we could, if we could have your.
David Taybron: Your point of view on this as well. Sam Navarro: Sure, and i'll touch a little bit more on the cooperative purchasing agreement I think again it gets abbreviated eloquent job on the use cases and so. Sam Navarro: It regardless of the space you're in within the public sector, the cooperative purchasing a great program opens doors for.
Sam Navarro: The Federal Government to collaborate and cooperate with state and local entities and in this particular area we open the schedules program to state a local entities to leverage what traditionally has been known as 70 schedule 70 and schedule it for now Dave was. Sam Navarro: Key in consolidating the schedule, which was a great program and effort to do so, those categories, no longer exists or what are they today it's just anything within the realm of IT technology or security, law enforcement. Sam Navarro: capabilities that are already on the schedule and So what does this mean we've done a lot of the work for you up front, we negotiated pricing terms and conditions. Sam Navarro: All of the security things we talked we just talked about are already incorporated. Sam Navarro: And also contracting officers, a specialist have done the diligent tasks of making sure that there's industry partners on schedule that actually have experienced that expertise within the area you're looking for so.
Sam Navarro: i'll just touch on two key topics that I think add value to the cooperative purchasing program one is bringing leadership to the table. Sam Navarro: So is there, someone in your organization asking the right questions as Eric and Sabrina said right, what are the problems we're trying to solve. Sam Navarro: And then, a different look at innovation right, I think, innovation, used to be, we used to bet the farm house on something and throw everything at the kitchen sink into one thing. Sam Navarro: And if he succeeded great you know you had the new Facebook sort of speak right, and if it did it.
Sam Navarro: You kind of ended up in the column of blockbuster right well that's that innovation anymore, I think, innovation, now is making small investments. Sam Navarro: In challenges you're trying to solve, and hopefully if they're unsuccessful they fail fast right but failing fast means you're learning faster. Sam Navarro: And the things that hopefully our successes, then you can scale on you can invest more time and effort and resources into. Sam Navarro: So I think How does that tie into the cooperative purchasing agreement access to industry partners that have experience in deploying solutions into live environments at a great price. Sam Navarro: Someone in your organization who's innovative and kind of Islam great leader to get behind and then make small investments into areas that you think can show promise and then.
Sam Navarro: focus on the ones that are successful so we'd like to be a partner with state and local entities in that journey we think we could provide a lot of value so after this conversation feel free to reach out to us we'd love to help you in your next deployments of Ai technology. David Taybron: Very good, very. David Taybron: Good so so that was the last question i've had, and this has been a very enlightening and great discussion. David Taybron: With each one of our our panel Members here, and we really encourage everyone in our audience right now to to use to chat any questions that they have and we'll take some time and answer some of those questions. David Taybron: We may have had some that have come in already but we're really encouraging our audience to take advantage of this time and GSA and our community of practice and.
David Taybron: In our centers of excellence here we're able to afford you the opportunity to to to ask questions directly of us, and if we obviously can't answer will get you resources that that can so. David Taybron: But but but anyway, David i'd like to kind of turn it back over to you, but I do really want to thank Sam Sabrina and Eric for for your time today and your your responses to all of our all of the questions that we had and and. David Taybron: I want you guys to take a virtual bow before we start taking questions from our audience so so again, thank you for your for your time here. David Taybron: and David i'd like to turn it back over to you, sir. David Orcutt: Sure yeah thanks David and thanks to all our panelists very enlightening discussion we had a few questions that came in in the Q amp a pod and. David Orcutt: it's not too late to type in a question, if you have one folks just click that little Q amp a button at the bottom of your screen.
David Orcutt: We might be able to tackle one or two, but no worries we're going to catalog all these questions and post them to. David Orcutt: interact that gsa.gov and I can send that a link to you and we're going to actually produce this recording and put it on YouTube. David Orcutt: as well, but i'll give you materials tomorrow to all of our attendees so no worries there and you'll get one CFP continuous starting point credit as well, so the first, the first question, I see in the in the Q amp a. David Orcutt: The person asked, I think, Google is collecting data to be able to send me advertisements that might be interested in I think we've all experienced that.
David Orcutt: For example, when I searched on one term one day I might see an ad for something related on the next search does Google send reports on what is most searched on for any GSA products in contracts, I think, maybe Sam you, you were starting to type an answer for that one. Sam Navarro: For for in regards to Google. David Orcutt: yeah in regards to Google ads. Sam Navarro: yeah I think. Sam Navarro: I think they made it evident, and you know very interesting, I have a buddy of mine darryl people works for Google and I was telling him that quick use case so quick answer is yes, they do that but. Sam Navarro: Every 90 days i'm always googling and I service department for my tucson to get the oil change and what recently has happened to me, is now I get oil change advertisements every 80 days to 60 days.
Sam Navarro: From Google right to get the best bang for my buck and I thought it was a buddy of mine who work for Google is trying to be up but it turns out, it was an algorithm. Sam Navarro: i'm being silly but. Sam Navarro: The answer is yes, as a way that works in the background is kind of. Sam Navarro: optimize our experience with some of these solutions. David Orcutt: Great Thank you yeah I know we're we're already coming up on time I can't believe in our just kind of flew by here but. David Orcutt: we're going to catalog all the questions that came in folks and you'll be able to see see the question with a with a well written answer we're going to share these with our subject matter experts so.
David Orcutt: And, and in the recording someone asked, will we be able to see the links that we're in the chat and the answer is yes, as well, so you're going to get all of those resources. David Orcutt: In an email that i'll send to you tomorrow, along with your CFP credits, so you know, I just want to again thank you, David for moderating Thank you Sam. David Orcutt: And Eric and Sabrina for the excellent advice that you gave us during this hour and I did put in the link, we have a training coming up June 8 and i'll send that as well. David Orcutt: that's an introduction to how you can use how state and local government can use our solutions here at GSA we have a wide variety.
David Orcutt: of solutions that you can use here, so I encourage folks to sign up for that training as well, and again just Thank you to everyone for your time today. David Orcutt: A real informative discussion on artificial intelligence, I hope we are able to produce more of these panel discussions in the future and its really, thank you for your time and your insight today, thank you to the two hall of you. David Orcutt: Thanks Eric Sam David Sabrina take care. Eric Ewing: Thank you. David Orcutt: Thank you to our attendees i'll send out all the materials after after tomorrow, and thank you for your time as well.
Sam Navarro: Thanks everyone. David Orcutt: Thanks all have a great day. David Orcutt: bye bye.