Skip to content

Business as Usual: Army Artificial Intelligence Task Force

Today on Business as Usual, learn about Pittsburgh's role in the U.S. Army's  Artificial Intelligence Task Force (AITF) from its Deputy Director, Colonel David Bradshaw.

AITF allows the Army to better connect with the broader artificial intelligence community and focus their efforts in this dynamic field. The Army Futures Command established the “AI Hub” as the charter location for the AITF.

As part of Army Futures Command, the AITF will leverage the AI Hub and engage universities and companies from across the nation by building upon initial successes at Carnegie Mellon University and the vibrant robotics and AI communities in Pittsburgh and Austin, Texas.

 

Transcription: 

So good afternoon, everyone. This is Audrey Russo, President and CEO, the Pittsburgh Technology Council thrilled to be here, we have another great guest today. And I'm going to formally introduce Colonel David Bradshaw in a moment. But before we get started, I just want to give some, you know, appreciation to those who have helped us along the way. And that includes Huntington bank, they've been our partners right from the beginning. And if you don't know them, you do get to know them from your business and personal banking, as well as them being one of the most active in in SBA activities. And we know that COVID has definitely put them at a higher throttle, but they have huge amount of civic engagement in the community, and cared deeply about the tech and innovation workforce. Also 40 by 80. That's the longitude and latitude of Pittsburgh. That's the wholly owned subsidiary of the Pittsburgh tech Council. And it's our charitable arm. And we work on things such as entrepreneurship, and workforce development. And soon you're going to hear more about our apprenticeship work. So pretty excited about the work that we've been doing almost a year each and every day at 12 noon. But it's not the only thing that we do each and every day. So there's lots of other connections and team meetings and collaboration and business development and public policy that we do all day long each and every day. So very thrilled to have Jonathan kersting, as always joining us and he is vice president of all things marketing media for the Pittsburgh tech Council, he will be keeping his eye on the chat. And just a couple of things about that the chat is not for you to sell your wares, the chat is for you to ask questions of our guests. And today, that's Colonel Bradshaw, you can sell your wares during many other offerings that we have. But today is really just to talk about the topic that we have at hand, and we've muted your microphone. So you know, we making sure that nothing in the background is going to disturb our guests. So I'm very thrilled to have Colonel David Bradshaw, Deputy Director of the US Army Artificial Intelligence Task Force, I had a chance to talk with him the other day, and I really think that you're all going to be in for a treat. I'm very thrilled that he is going to be based here in Pittsburgh. And I think all of you are going to be excited to just allow us to peel back the onion just a tad to get a sense of all that he's working on and leading. So I'm going to bring to the forefront. Colonel David Bradshaw. Welcome. Welcome today. And thank you, I know that you're not in Pittsburgh, you're traveling, and you made sure that you carve the time out for us deep appreciation. So welcome.

Thanks, Audrey, I appreciate the time to join the group and have a discussion about where the task force is and where we're headed in the future, kind of what projects we're working on and, and how we can link up with the local ecosystem.

Well, before we even jump into all that, because I know you're gonna have a lot to talk about, let's talk about you. Let's talk about your professional journey. So people get a little bit of an understanding of the kinds of things that you've been involved in and what you're passionate about.

Sure. So, way back a long time ago in 1991, that was commissioned as a second lieutenant out of Wentworth Military Academy, which was a small military junior college and then went on from there to get my bachelor's degree in math. So you know, foundation and mathematics, was a basic branch field artillery officer. And then in 1998, was offered an opportunity to go teach it the math department at West Point in New York at the United States Military Academy. So after leaving battery command, I went to grad school at Rensselaer Polytechnic Institute, in beautiful Troy, New York, spent two years there, got a master's degree in applied math and a master's degree in art and statistics, and then taught math for four years at West Point. After West Point, I actually ended up in a great assignment working counter improvised explosive devices. For the Department of Defense, really cut my teeth in data science. They're doing a lot of predictive analytics, pattern development, a lot of social network analytics to get after what we saw is one of the biggest problems facing troops in Iraq and Afghanistan. After after the joint ID defeat organization did a short stint at Human Resources command working army personnel issues Went back to the joint ID defeat organization for another tour, and then went to the trade off Analysis Center at White Sands Missile Range. And our focus there was really on analysis of alternatives, cost benefit analysis for army programs. And then from there, I went to Special Operations Command, where I was the director of analysis and strategic planning, and also served as the first command data scientist at at SOCOM and stood up the brand new data engineering lab there in Tampa. So soft works, folks, we were co located with the software folks or soft works folks there in Tampa and then just left there this past summer. And I've been in in Pittsburgh since September. We're now I replaced then, Colonel Maddy. Now Dr. Maddy, he was the deputy at the task force before I was so Doug and I have had a long career, we actually taught math together at West Point. So we've found a parallel careers. Although not serving together, you know, we've been working side by side for a pretty long time.

Wow, that is just more than impressive. You do have big shoes to fill. But based on your background, I think that's going to be pretty easy to do. So we're thrilled to have you here in Pittsburgh. And and let me talk about now the army AI Task Force and the role that you have. And then we're going to talk about, you know, what the task force does, what's new with the efforts, you know, the priorities, etc.

Yeah, so So I would say it's probably not lost on anyone on the call that you know that the government as a whole, not just God, but the US government as a whole understands and is starting to realize the value of data, the amount of data that we're actually producing on any given day. And then if you take that data and couple it with now the compute power that's available, the ability to transport mass amounts of data fairly, fairly easily. Cloud Compute Cloud Storage, I think that what you're seeing is the department is is understanding that there's value in that data, and there's decisions that can be made both to be more efficient and more effective. And what we haven't done really, as a department is wrap our arms around it in a in a cohesive manner. So I think what you're seeing is that the department recognized the issue stood up a task force in the AI Task Force, and really, the focus is for us to start getting after those problems, whether it's integration, whether it's development, to start, start down that path of developing an AI solution or AI foundation for the Department of the Army, and also partnering with the Department of Defense, in a broader sense.

So if I recall from our conversations, right, we you talked about that there's really four pillars or four, you know, effort to focus, can you can you talk about those focus areas, and then maybe we can dive into a few of them?

Sure. So um, you know, when the, when the task force was stood up, it was really a mandate from then general McConville as the, you know, Chief of Staff of the Army and Secretary McCarthy when he was the SEC army. And, and their, their focus was really, you know, as, as you want to build synergies around artificial intelligence, you have to have some quick wins, just to gain some momentum. And what they really did was focus us on four areas. So one is human resources and talent management. You know, we as the army see sea personnel and people as our number one priority, you know, without the people you don't have an army so so general McConville direction to us was Hey, make HR talent management the number one priority it's really the the age old problem we have in the army is you want to have the right officer or or soldier right soldier with the right skill set in the right place at the right time. And in years gone by you know, it was really a numbers problem for the army. It was, I have, I have open billets, I have people with skill sets. First In First Out, you know with with some pretty rudimentary rules. That's how we how we move talent around the army. It really wasn't talent management, it was really about personnel, moves and personnel, permanent changes of station. So what we're trying to do is, is peel that back a little bit, you know, and our thought would be, create a site similar to, you know, maybe a monster.com, or name, your, you know, name your employment app. And it's really about understanding what skill sets people have, what skill sets are needed in the army, and then aligning those skill sets better across the army. So that we get a more robust army with better, better capability. In the end, you have a more robust skill set from unit to unit. So it's really a two part problem. Understanding the skills that people have and then understanding the skills that are needed from the unit's perspective, and then meeting in the middle to marry those up. So we definitely see HR talent management as one of our focus areas. Predictive Maintenance is another another great area for us. There's, there was some work done last year in partnership with Carnegie Mellon on the you h 60. Blackhawk helicopter, for one of the special mission units in Special Operations Command. And it was really about how you can predict and determine in an engine failure before it happens. So you know, the 160 at soar which is special operations, aviation is a very small unit. So it's pretty easy to to build a model understand all their data, and capture all that data in a single place. What we're trying to do with RPM max effort is expand that to the broader Army Community. So how do you take the very small niche effort that was done with the one 60th and expand it to the broader army Blackhawk fleet? And then from there, if you can gain winds off of that, how do you expand it to other air platforms, helicopters and fixed wing. And then if you can do that, you can you can get into ground ground vehicles, you know, Humvees tanks spread lace. So so we're not trying to boil the ocean. But we are trying to, you know, have some have some individual niche winds that we can expand on as an enterprise to, to build that momentum and help people understand what data is available, what data is needed, and how you really get after artificial intelligence at scale, for a logistics program, the other one would be aided threat recognition. So, you know, there's been a whole lot of work done on UAVs, and the ability to use computer vision and artificial intelligence to identify, you know, trucks, people's buildings. We are taking it from from a different perspective. So we have quad copters and ground vehicles that are working as teams to identify targets or identify,

you know, things of interest. I won't say targets because not everything's a target but but items of interest using computer vision. So in partnership with Carnegie Mellon and the National robotics Engineering Center, you know, we we have done some work to develop a MVP, I would say that we deployed out to project convergence last year, which is project convergence is the the capstone event basically for army futures command every year where you can go out, test your test your goods, test your wares, and then make some determination on how you want to move forward with those projects. So last year, we took two vehicles and two quad copters this year we'll probably take four quad copters and for military vehicles last year we use john deere gators this year we we expect to use military grade vehicles in partnership with CMU we reached out to be a kinetic Textron you know kinetic being a local company is really good for the ecosystem. So we so we're looking to you know, almost double in size, what we took to project in virgin 20 in project and Virgin 21. So, really excited about that we see some some great movement on that project. In the future. Let

me ask good things and we're gonna we'll have some, I think there's a couple of questions. Where are you going to be located? in Pittsburgh? Are you gonna have your location and then we're and then Jonathan's gonna talk a couple of the questions.

Sure. That's a great question. So So when the taskforce was stood up, we initially were located at the National robotics Engineering Center right off the 40th. Street Bridge. Right. So in Lawrenceville, yeah. So we have some space carved out there. We also have space in the ghatak building. And I'd be lying to you if I told you what to get would get arrested for but it's on the CMU campus. It used to be I believe it was a Microsoft development space. All right. So So we have, you know, between those two spaces, we have roughly, you know, Teach For about 50. Folks. With our current plan, we will quickly outgrow that space. So I think you know, from Dr. Maddox perspective, we look at the the mill 19 area and some of the work. Yeah, so Hazelwood green area, and that area that's being expanded in in our mind, we think that's a great fit for expansion into that space, given the the resurgence in that area. So so I think you'll see me know some movement with the task force. As far as space goes, we just haven't got a finalized picture on what it's gonna look like in the end. But right now we're in two spaces. So the national robotics Engineering Center and the data building on the campus, people working remote. So we we have a hybrid approach approach right now. So we have, you know, folks, at n rec, and we have folks, soon to be in Ghana, we just got that space. So so we're working to move into that space. But we've basically taken the task force and broken it up into a multitude of teams. And then based on the work that's going on, on any given day, we rotate teams in and out.

That's great. So let Jonathan, just grab a couple of questions. And you know, there's a few, so let's sort of go through them.

Rochelle, thanks for being here today. This is a very fascinating conversation. Just love it. Starting at the top here, just curious to know, are you using AI to, you know, match God needs to appropriate personnel that already in the military or to recruit new people to the military?

So a little bit of both, I would say we just kicked off. So. So the short answer is yes, we are currently looking at personnel that are already inside the army. So that would be that that matching problem we have? So so we're doing that right now, we just kicked off a project with the United States Army recruiting command, to help them look at how to better more efficiently, more effectively recruit people from inside the US, right? So how do you go out? And you know, rather than shotgun blasting your recruiting dollars across the entire United States? Does it make sense to have a more refined recruitment effort in a certain area after a certain demographic after a certain, you know, skill set that we're looking for? So so we're trying to work a project with user rec right now? On on getting after just that problem? So it's really, it's both both aspects. So those that are coming into the army and those that are already in the army.

Excellent. And also, so I'm in your work? Is the army AI Task Force looking at individual or team performance data? And how to leverage that, and what are you seeing so far?

So I, so the army individual performance data is, is really, I guess my question would be, are we talking evaluation reports? Are we talking physical fitness? I'm not sure I fully understand the question, but I'll, I'll run with it a little bit. So I would say from a team perspective, you know, that's not how the army is, is really geared we look at individuals when we're talking about moving people. Now there is some some thoughts on how we build cohesive teams. You know, from our perspective, how do you build a data science team? And I use the term data science loosely because I know it's it's not real well defined, but, you know, how do you build a data science team that you can move as an organization throughout the army to to be effective and efficient From a personnel side, you know, we're really limited on the type of data and and kind of data we can grab about a person, most of its from evaluation reports. And evaluation reports right now aren't in a format, or form that would allow us to really grab quality AI, you know, value data that we could, that we could really build AI off of, we're looking at some options as to how we might do that. But I think what we'll find is, we need to reconfigure how we're evaluating people's skill sets.

Here's a really good question, I think rolling. So to learn more about so what is the difference between the army AI task force in the army artificial intelligence Integration Center? And how do your efforts differ?

So, so those are one in the same. So so the AI Task Force, so in the army and Department of Defense, the task force really has the shelf life, right. So you stand up a test a task force, for a limited duration time. And then if you want to make it permanent, you either have to change it to a center, make it a regular, regular unit, or you you build it as an organization underneath a different center organization. So what you see is the Army's artificial intelligence Task Force is actually converting into the Army's artificial intelligence Integration Center. Really a lot of the same mission sets. But our focus as we expand, will be more on the the integration side of the house. So how do you wrap your arms around everything the army is doing, and try and help coordinate efforts from unit to unit and command to command?

Okay. And then one more question from LA.

Yeah, it's a great question. Belize wants to know, here. So what funding opportunities are available for new technology development for small businesses?

Sure, so there's so we we go about it in a couple of ways. So clearly, there's the Department of Defense and, you know, the acquisition side of the house, broad area announcements, which will we're working on right now, some some broad area announcement language that will be published and released to the public. And that allows companies and organizations to send us white papers and thought pieces on where they might be able to answer questions from a broad perspective. And then the other option is commercial solutions opportunities, which is just another another acquisition vehicle that allows folks to provide input to specific questions we have about technologies and and what other companies are working on in in specific spaces. So I think Lindsey and Jane from my team can can drop some links into the chat. If folks are more interested in that, and then I can link or we can link them up with our financial and business office for industry engagement type stuff.

Just one more question, Jonathan, you want to grab that brandy? And then

I love it, man. The questions keep rolling in this is good stuff. So Randy wants to know, how do you see artificial intelligence and machine learning being implemented and deployed large scale into the army in God? And more importantly, how do you plan to test the accuracy of those platforms?

So So I will tell you that is that that is a great question. And I think God is struggling, not I won't say struggling is, is exploring that right now. So my my question is always, if you build generalized AI, which is great, say you're, you know, you build computer vision algorithm to identify a truck. And then you want to take that algorithm and move it into a different landscape. So if you've, if you've tuned your AI to look for a truck in a desert, how do you then take that AI and tune it to look for a truck in a in a forest? The intent is that people would be able to tune their own algorithms. I don't think the army has has really nor God thought through the policies and permissions that allow folks to do that at the tactical edge. So I think we still have some work to do in that space. I think we have some very niche AI solutions. But you know, for from the task force perspective, we like to talk about an AI stack, which is really generalizable across all solutions. And you know, that's not unique to us, we really took that from Carnegie Mellon, as they talk to us about AI and what underlays, all of the AI stack is AI ethics. And we are really concerned about the ethical attributes that go with every stage of artificial intelligence development, from from data creation, to data display, you know, and all the compute transport and algorithm development along the way. So, we do have an ethics officer on the team. He's currently teaching at West Point. So we take ethics very seriously. And, you know, I think we still have some work to do as far as how you generalize AI and do it at scale at Echelon in the army. But you know, that that's one that I think the task force in that when we change into the center, will definitely be taking on as a huge undertaking.

So we only have a few minutes left. And I think there's one piece that I want to make sure that you can give some examples of and you and I talked about this, like what give us a great example of industry partnering with your team?

Sure. So, you know, I talked about my background, I'm an operations research systems analyst, I have been for about 20 years. And I will tell you, from the army perspective, as a data scientist, to work on VOD networks is kind of difficult. We've never really had a data science development platform available for us. And we've known that as a problem as a branch for for a while. In August, we undertook the development of a data science AI, you know, Dev sec ops type platform with visit Mo, which is a company there and in Pittsburgh.

Last week,

yep. So so we undertook the development of this platform, starting in August. And in rapid time, in December, we were able to produce a minimally viable product, that we were able to ship out to a few folks and have them test drive. We're currently refining the product. But you know, to go from, from concept and idea to deployment and development, even with an MVP, in about four months time is near impossible inside the Department of Defense. So So I would say that is a huge benefit to us is to be able to reach out to the local community, you know, on the AI side of the house, on the software and tech development side of the house. And in the robotic side of the house, there's just so much that Pittsburgh brings to the team, that it's great for us to link to them, either directly or or through and with CMU. So some great opportunities. What was the product? Can

you mention what the product was not sure it was called

co ees. co E, US Eus. CO is a Greek want to say great Titan of intelligence?

Yeah. Okay. So that's great. So is there anything else that you think that we should know? We're wrapping up the show, and you can see people have questions, people are excited, we're thrilled to have you in Pittsburgh, and excited that you're going to have hybrid models, and you're going to be dispersed around the city. Really important work. Is there anything else? Is there something next that you you want to mention? I mean, I know you have these four pillars, and you know, you're very intentional about the work. So,

so So I would say, you know, hit if Jamie and Lindsay put the link in the chat, please, please reach out to us via the link, watch, watch social media watch LinkedIn. You know, we'll we'll have some announcements via those those platforms. And you know, we'll be transitioning into the center. Hopefully, sometime, we're thinking mid April. So, so please watch for that. But by all means, reach out to us. We're always looking for willing and unable partners on the AI and tech front.

Well, Colonel Bradshaw, thank you so much for dialing in today and joining us, we will make sure that people have your information someone's asking about LinkedIn, confirmation, contact info, everything's out there in the chat. So you should be able to reach them. It's not just reach out to Jonathan or myself or Brian and we will figure out ways to stay connected. This is the only the beginning of the conversations that we're gonna have with this team. So we're Pretty thrilled about being able to just sort of shine the light on that today and tomorrow, but if you stay tuned for tomorrow, I know that we gave a plug out here about the innovations in unconventionals. And that's coming up in like three minutes. So if there's anything that you want to join us in terms of Pennsylvania oil and gas in the clean tech form, we're doing that in just a moment. And tomorrow, we are going to welcome some new neighbors on the north side that are part of a company called parallel, and it's called good blend. And good blend is a new marijuana dispensary. And they are here in Pennsylvania and a few other states obviously in Pittsburgh. So it'd be pretty interesting to talk about what their plans are. And so I want to thank everyone in particular Colonel Bradshaw and team. We look forward to partnering with all of you and see you here tomorrow, same time. Really appreciate it.

Transcribed by https://otter.ai