How can the Federal government get in front of two of the dominant – and in some measures alarming – forces of change on the technology landscape and set a course to harness the potential of those trends for improving service delivery and customer experience?
We snagged 30 minutes on the phone with the General Services Administration’s (GSA) long-time CIO David Shive and came away a whole lot wiser about how not only GSA – but the entire Federal enterprise that it serves – can manage both the unavoidable challenge of putting artificial intelligence (AI) technologies to work, and also how to evolve the government tech workforce to meet the demands of the next evolution in tech thinking and practices.
For those who may not be familiar, Shive’s thoughts on those two issues carry with them the weight of a unique set of experiences running technology within the Federal civilian enterprise. He’s approaching his ninth year as CIO at GSA – spanning three presidential administrations – is co-chair of the Federal CIO Council, a permanent board member of the Technology Modernization Fund, and a FedRAMP executive board member.
With the benefit of that experience, here’s what Shive is thinking about embracing emerging tech, and how to re-vision the tech workforce to meet the moment.
MeriTalk: We know that GSA is always working on technology projects at many, many levels, both internally and for the rest of the government. Can you tell us a little bit about something in the works on the “cool tech” front?
Shive: The cool tech that we are working on now actually has to do with people. I learned early on in my career that tech done for tech’s sake doesn’t provide a whole lot of value. But tech done in a way that serves the needs of people – that provides the greatest value.
Just about everything we do here in GSA IT – my shop – in support of the agency’s business mission is really all about people. When I take a look at the disruptors that are in the marketplace that might affect and change people, AI is central to that view.
AI has been there for a while, and generative AI is quite different than the AI that’s been in use. So I am looking at how I need to focus my IT shop to be ready for the benefits and values of things like generative AI.
Over the last year or so, when I worked with my executive team, one of the things that we’ve been talking about as we deliver IT and support the GSA business mission is what changes do we need to do to ready our workforce to be able to work in the advent of augmented capabilities for delivering IT.
I used to say we have to be ready because within 10 years from now, we will be speaking code into existence. I’ve had to change that quite dramatically. Now when I talk with my executive team, I say we need to be ready to speak digital business capability – not code but digital business capability – into existence in the next three years. So what we would be speaking into existence has changed, and the timescales have changed pretty dramatically.
MeriTalk: How is that mindset shift changing things?
Shive: This has impacts on multiple fronts. I run a fully consolidated IT shop. We did that consolidation play eight or nine years ago where we took the IT shops that were spread all throughout GSA and consolidated them into one IT shop. As a part of that we ferreted out shadow IT that would try to stand up, things like that.
In this new world, one of the ways I’m getting my team ready to deliver is recognizing that there’s massive technical capacity out in the businesses of GSA, and as a top-level tech executive, it would be foolish to ignore that capacity. So when I say we’re readying ourselves to speak digital business capability into existence, that’s not just my IT people. I’m looking holistically at that capacity that exists all throughout GSA and asking what do I need to do to ready all of those tech-aware, tech-capable, and tech-adjacent people to be able to do this very thing.
MeriTalk: So in other words, remaking the definition of tech workforce to some degree?
Shive: Absolutely. Organizations need to continuously evaluate how their teams are working together, delivering, and staying on top of new technology trends and skills.
It’s going to mean that they’re talking less at the assembly language level or the third-gen or fourth-gen computer language level, and they’re going to be elevated up into understanding the business well enough to be able to say I need a system that does X, Y, or Z. And then take a look at what’s generated by that ask and say did this meet the mark, where should I apply human intelligence to the AI-generated capability to refine it, make it better, make it the best thing that it could be, make sure it’s very tightly coupled with and aligned with the business need as it exists there at the moment. And then be able to shape and modify that in near-real time as business priorities change as business needs change, and as enabling technology changes.
MeriTalk: Generative AI is supposed to get journalists worried because you can just ask the AI to tell you a news story. Is there a similarity on the coding front where you tell the AI what you want to accomplish and it will give you back a technology map that you can then evaluate, with the effect of taking away the sorts of conversations that happen now only in coding language?
Shive: Yes, to a degree. When you look at the outputs generated by machine learning and AI right now, they all have biases introduced by the original developer of the models that make up those capabilities, but the thinking is that those things could self-learn and self-correct over time, especially with additional inputs.
When we talk about a request to build me a technical architecture for X, Y, or Z, those biases are resident and that’s why you’ll hear me talking constantly about the value of augmentation through AI. But it has to also be augmented by what I call biological intelligence – that means humans. Because they are the correcting agents for the flaws that are inherent within artificial intelligence. We have to build the control environments to be able to make sure that biological intelligence is always there to be the guard rails for artificial intelligence and machine learning.
MeriTalk: Is there a timeline for when you and GSA started going down this path?
Shive: About a year ago, we took a look at the software development lifecycle, which has been shortening our time to value in some of our business operating units. You don’t get those time-to-value levels without significant automation and augmentation.
We can no longer look at artificial intelligence as an augmentation, we have to look at it as a primary delivery capability. And to make it a primary delivery capability, we have to make sure of three things. First, that you have a viable strategy for it. Number two, that you have viable control environments around that thing to safeguard and protect the interests of the government, people, and all those who might consume the outputs of those things. And number three, make sure that you’re well positioned and well prepared as an enterprise to live and react in that new world. That happened about a year ago.
MeriTalk: If that was about a year ago, then the new thinking from it is still fairly new. Since then, what’s been going well with the process?
Shive: What’s going well is we have the right people looking at this – both the right techies and the right business people.
GSA has a pretty significant outward facing mission delivering value and savings in real estate and acquisition and tech – both in the tech delivery side and the tech policy side. So we’ve taken kind of a holistic team view-based approach to this. That’s always good whenever you have disruptive technology.
The other thing that’s good is we are trying to stay as open-ended as possible, and saying we’re not going to know all the values that can be generated here, and that we’re going to learn about the value propositions as we move through this. My job is to make sure my team is prepared, make sure that there’s an appropriate control environment, make sure that the legal and ethical implications are put in place and that those are not restrictive to that ability to generate unknown values. That’s working pretty well.
We’re putting kind of a U-shaped control around this and leaving an open end on the top for the things that we learn along the way. Those can funnel into those control environments and where we need to we can modify control environments, modify how we’re training and teaching our techies to deliver in this future, and modify how our businesses are going to consume the outputs and value generated by these technologies. That approach is toward all of our technology development.
Here in GSA and specifically within GSA IT, we haven’t had to change how we assess new and enabling technologies as they present themselves, our ability to ingest those, assess those, derive value out of those, and then expose those at scale through our product and service offerings. That whole mechanism has worked really well here – it’s been in place starting with mobile enabling the Federal government, or using technology to change business processes. We’ve used this kind of same process and we’re finding that it works really well with AI and generative AI and machine learning.
MeriTalk: How big is the GSA IT group, roughly how many employees?
Shive: We have about 500 Federal government full-time equivalents, and then contract staff – that can expand or decrease. GSA as a whole is a big business enterprise. We’re the biggest property manager in the world, we’re the third biggest design and construction company in North America, we’re one of the biggest travel agencies in the world, and one of the biggest fleet managers in the world.
MeriTalk: Along with what’s going well, are there things you are identifying as challenges?
Shive: This represents a fundamental shift in our approach to delivering technology and that type of disruption can be challenging on two fronts – change is hard, change at scale is even harder. So that’s something that we have to be thoughtful and honorable about.
I’m responsible for 500 Federal government employees. I have an obligation to make sure that their careers and the things they’re working on remain relevant, and how they deliver to make sure they’re trained well to operate and deliver in this new model. So that’s challenging – to shift an entire organization to a new way of delivering.
It’s also challenging as an organization consuming the IT that I deliver to our businesses if they have to consume tech in a whole new way, that can be challenging. We as humans can be resistant to change. I fall in that category and look at me – I’m an IT person.
Whenever there’s disruptive technology, there are great gains potentially, but you also have to be very thoughtful about how quickly you pivot an organization. Here’s why: We do tons of work in this government, and for our government partners, and their need to consume the services that GSA generates isn’t going to change regardless of the fact that there might be disruptive technology presenting itself. Their needs are going to continue, and we have to continue to provide a very high level of service and product while also changing beneath the covers. That’s not easy to do. That’s a risky thing to do, but every once in a while, it’s the right thing to do.
We here at GSA don’t really stand pat very often. We’re always looking to say what’s the next thing that we should be looking at to operate more effectively, more efficiently, at a higher velocity, and at higher quality on behalf of our agency partners. That’s tough, but that’s why executives exist – to play that balance, to continue to generate value while also transforming underneath.
MeriTalk: Is there anyone in government or elsewhere that you’d like to shout out for good collaboration as GSA undertakes the kind of shift you’ve been talking about, perhaps on the Federal CIO Council where you are an influential voice?
Shive: I’ll single out the CIO Council first, because it’s easy. Federal CIO Claire Martorana and I have the great honor of co-chairing the council, and a lot of the thinking that I’ve been doing here in GSA IT is the same as many of the other CIOs have been doing as well. We have frank, open, and honest conversations about all of this in that forum. Good Fed-tech demands that we be talking about this kind of stuff and thinking about this kind of stuff. It’s great to have a forum to talk to my peer CIOs and gain their years and years of expertise and talent and apply that to what I’m doing there. The CIO Council is a great place to be having those conversations, coupled with the Federal CISO Council so we can be thinking about cybersecurity implications, coupled with the Federal Privacy and CTO councils.
It’s a really great ecosystem to have those conversations. It makes it so that none of us are headed down the wrong track too far. We can kind of self-correct each other as a community. That’s really great.
On the private sector side, there are some organizations that are doing some very creative thinking here. I won’t name any by name, but what I will say is that a lot of the vendors that are making low code/no code environments are doing some very creative thinking. It’s clear that they’ve assessed that the way that business capacity is delivered is changing, and they are taking a look at ways where they can jump in early and often.
That’s going to benefit the government because the government isn’t going to have to build this stuff from scratch. You can work with trusted partners who have done some deep thinking and been in this space for a while, and who can say here are tools that are cognizant of how delivering tech might change, but also cognizant of the specific needs of government.
MeriTalk: Do you have any advice for other Federal technologists on how to approach the same path as GSA IT is taking?
Shive: It’s fun to work on whiz-bang technology, but you have to draw a straight line between the thing you’re considering and good user outcomes and good customer experiences. So keep user experience and customer experience at the forefront of every assessment of new tech, and new tech process, and refactoring the way you work. If you can’t draw that connection, one of two things is going wrong: either there is no connection and you shouldn’t be doing the work and shouldn’t be wasting the time, or you haven’t fully thought through the thing that you’re trying, and you don’t have a big enough and deep enough understanding about what the impact is going to be on CX and UX and you should pour some more time and attention into that.
The second thing is to be strategic. There’s no such thing as a 100 percent value proposition, and there’s also risks associated, and we’re obligated to manage to both. Not having a good idea about both means you are operating at your own peril.
The last thing is to continue to curate a culture of change. It’s hard not to in the tech business, because our business is a business of change. But there are certain peaks in change. I suspect we’re approaching one with generative AI, and recognizing that, and curating your culture, and maintaining your culture through this time of rapid change is pretty important.
MeriTalk: Let me loop back around to the idea of reshaping the workforce by wrapping in tech-adjacent folks more tightly into the larger effort. Do you see GSA’s path – with some increasing embrace of the benefits of AI and getting away from being so code-heavy – helping to meet the government’s tech workforce needs?
Shive: I suspect yes, and that’s one of the reasons why we’ve headed in this direction. There’s more work than there is available resources right now, and there is also more capacity than there ever has been in technical capability. You see people with their MBA, MPA, MPP, but they have a minor in tech, or a computer information systems minor. All of that is untapped talent, and we would be fools to not find elegant ways to use that talent.
MeriTalk: Let’s close with something on the more fun side – tell us a little bit about your own path to a technology career. Has tech always been a natural interest for you, or was it something you picked up along the way?
Shive: It’s always been a natural interest from the first computer I had – an Atari 800 – where I kind of taught myself BASIC. I’m dating myself there. My first couple of jobs I had working my way through school included one where I talked my way into a job writing machine language-based drivers for big industrial manufacturing computers, stuff like that. It’s always been kind of a natural fit for me.
MeriTalk: And then finally, is there something you like to do in “real” life that has nothing to do with technology at all?
Shive: The easy answer is I like to run, that’s my preferred way of exercising. But the better answer is I have a whole life outside of Fed tech. I’ve been a foster parent for 20 years. We’ve had 38 kids come through the house, and my wife and I have bent towards what they call transitional care, which is providing a safety net for kids that age out of the system.