U.S. General Services Administration
Remarks for FedScoop AITalks on April 18, 2024 Prepared for Administrator Robin Carnahan
Thank you, Goldy. I’m glad to be here with folks across government and industry committed to delivering safe, secure and effective technology for the people we serve.
When I think about the world of possibilities opening up before us with the use of Generative AI, it reminds me of a whitewater rafting trip I made several years ago on the Colorado River through the Grand Canyon.
If you’ve spent time on a river, you know that trying to paddle against the flow is exhausting and often futile. The smarter approach is to take a beat, assess the terrain ahead, and look for a safe line through the rapids. Often there are a few different lines, but once you pick a line, you’ve got to get everyone in the boat rowing in the same direction to hit V at just the right spot.
By the time we approached Lava Falls – the monster Level 10 rapids on the Colorado – we’d already been on the river for several days, so we knew it was time to stop, examine the landscape, and scout our lines. We quickly spotted the riskiest boulders and holes and chose routes to steer clear of them. Then, we hopped back in our rafts and went for it.
One group hit their V perfectly, made it through fast and popped out the other side… wet, but exhilarated.
Others missed their lines and paddled frantically to avoid being sucked into deep holes…
And another raft missed their line completely, got hit by a giant wave, and ½ the people were thrown from the raft while the guide scrambled to pull everyone back to safety and navigate backwards through the tumultuous rapids.
It strikes me that just like that mighty river, our AI future is moving fast. There’s no stopping it, but, like folks rafting that river, we can learn to navigate it and even harness its power.
But, to do that, we too, need to take a beat and examine the landscape, identify the biggest risks and work together to guide our organizations forward to our destination – which is to deliver exponentially better service to the American people we serve.
Let me see a show of hands…how many people here are from federal, state, or local government?
It’s great to see so many of you. You should know that our GSA team wants to be your trusted guide to help you navigate this moment of change. We want to help set your teams up for success by providing the tools and skills you need to meet these new challenges and navigate this uncertainty in ways that are safe, secure, and effective for your organizations.
Today, I’m going to talk about how we’re doing that.
First I want to remind everyone that government is – at its core – a service delivery business.
These days, people expect online services and digital products to be simple, personalized, and easy to use – and the latest Generative AI technologies will only elevate those expectations.
That’s why it’s never been more important for government to deploy technology tools that not only meet those expectations, but do it in ways that also align with the values that we care about in a democracy.
So what are those values? (What is the sturdy, thick, stable material that our whitewater rafts are made of?)
When it comes to deploying Generative AI technology those values can be summed up in 3 words, secure, accessible, and responsible. Feel free to write those down, because these will be our guideposts as we work to deploy new AI products and services.
When I say “secure,” it means we’re focused on protecting people, their privacy, and their data… as well as our processes, systems, and national security.
When I say “accessible,” it means that, in a democracy, it’s our job to serve everyone, so the technologies and tools we offer have to be equitable, easy to use, and work for everyone.
And “responsible,” because when government uses emerging technologies, those deployments must be not just ethical, legal, but also fully transparent and accountable about the data sources, data quality and data models we use.
The President has been clear about those values, especially in his EO on safe, secure, and trustworthy AI.
With those guardrails in mind, I want to turn for a few minutes to how we intend to apply those and take advantage of the tremendous opportunities before us and meet this moment.
GSA has some specific jobs when it comes to technology - we build, we buy, and we bring people together.
First, we’re building AI capabilities within government.
We’re using our own teams to pilot new Generative AI services and explore use cases that can be shared and scaled across government. In fact, we’re already running 150 pilots using 132 different Generative AI tools and we’ve been using other machine learning and large scale data analytics and natural language processing bots for our call centers for several years. As of a few weeks ago, we have 7 sandbox environments for testing with more to come.
We’re out front on this because it’s GSA’s job to build shared services that can be reused across agencies and save taxpayer money. And we know Generative AI tools can help accelerate that.
Speaking of shared services, let me take a minute to update you on what’s happening at Login.gov – the single-sign-on account for all of government – because it’s an uniquely important shared service and a good example of how we’re using technology ethically, responsibly, and securely for the public good.
As you know, Login has been around for several years. It’s now used by over 40 federal and state agencies and over 50 million Americans – and that means more secure, easier to access government benefits and services for millions of families, veterans, farmers, students, and more.
We just announced that Login.gov will soon begin piloting facial matching technology to allow people to remotely verify their identities – something that Americans increasingly expect to use. And we’re doing it in a way that’s consistent with our values and commitments to privacy and accessibility. That includes minimizing data retention and ensuring that all personal information that users share with Login.gov can’t be sold.
Even with those commitments, we know that some folks still aren’t comfortable with AI-powered tech, so we’ve taken the step of partnering with the Postal Service to provide a visible, upfront, and accessible option to verify their identity in-person at one of more than 18,000 post office locations – and about 99% of Americans live within 10 miles of one of those locations.
In addition to that, we’re also wrapping up an equity study to help ensure remote identity verification services offered by government agencies serve all Americans effectively.
The lesson here is that we can be thoughtful and inclusive in how we build government-owned products and services and that includes Generative AI. Taxpayers deserve nothing less.
Second, we’re going to help agencies buy best-in-class technologies.
Partnering closely with industry will be critically important to helping advance our commitment to secure, accessible, and responsible AI.
That’s why our IT Category Office is developing an Acquisition Resource Guide for Generative AI and Specialized Computing Infrastructure. This is a big deal because it will help procurement teams across the government be better informed when they go to buy AI capabilities and computing infrastructure.
Here’s a sneak peek of what you can expect from that guide, which is coming soon…
- It will help agencies identify common challenges and use cases where deploying AI tools can have the biggest impact.
- It’ll help agencies navigate the marketplace and find the right solutions to meet their mission – and it’ll help them determine the right time to buy, or maybe when to push pause on a buy.
- It’ll also help agencies with market research on products and services that are already available or ready to be cleared through our FedRAMP program…
How many of you are familiar with FedRAMP?
It’s the first security gate that all cloud products and services must pass before they can be used in government. I’m guessing many of you have strong opinions about what you like and what you don’t like about the FedRAMP program.
I want you to know… we’ve heard your feedback and the team has worked hard to integrate that into a new Roadmap for FedRAMP. It’s designed to make the program more scalable, secure, and easy to use. I’d encourage you to take a look at that Roadmap and let us know if you think we’re on the right track.
But that’s not all GSA is offering. We also know that budget cycles are long which creates another challenge and oftentimes means agencies need actual dollars to buy AI and emerging technologies. That’s why the Technology Modernization Fund is such an important resource because it’s the place agencies can go for incremental funding of new proposals outside your normal budget cycle.
And right now, the TMF has an open call for AI-related proposals. We expect to see a lot of interest from agencies looking for centralized AI infrastructure and platforms. And we’re eager to collaborate with teams across the government to solve common problems and create more opportunities for shared services. It’s a perfect chance to avoid the all-too-frequent pitfall we see in government where every agency ends up reinventing the wheel.
When it comes to AI services, we have the chance to do better and the TMF is the best place to start.
I’ve covered building and buying, but perhaps the most important thing for navigating any unpredictable rapids ahead are the people with you on the boat.
That’s why the third thing we’re doing is bringing people together.
That means bringing in new AI tech talent, upskilling the great public servants we already have, and working across government and industry to get everyone rowing in the same direction.
We know that technologists are interested in serving their country, and it’s our job to create pathways that make it easy for them to say YES to public service. Between our Presidential Innovation Fellows program, the U.S. Digital Corps, and the roles we have in our Technology Transformation Service and GSA IT, we’re bringing more tech talent into government than ever before. In fact, our call for applications for our AI-focused fellowships received more than 2,500 applications!
We’re also providing upskilling opportunities for our inhouse talent so they can learn and grow along with these new emerging fields. A great example of that is our AI Community of Practice’s partnership with Stanford University to deliver training to nearly 5,000 federal AI practitioners – and there’s more to come!
I’m curious, let me see a show of hands, have any of you taken any of those Stanford AI trainings? I know, I’ve learned a lot from them. You should give them a try.
TTS also has published a great AI Guide for Government – it’s constantly evolving and specifically focused on the application of AI for government and you can find it online.
By publicly sharing our people, trainings, and expertise, we know it helps our government partners be better prepared to deploy AI tools equitably, competently, and responsibly in ways that scale.
Just like on the river, the more guides and teams share their collective knowledge and work together, the more likely they all make it through the rapids and safely downstream.
Bottom line: we’re committed to getting more people into public service who are ready to create and use Generative AI and emerging technologies for the public good. If you know anyone motivated to do that, I hope you’ll send them our way.
So, what’s next? What direction should we be rowing?
Well, for government partners, our agencies need to work more like one government. That means finding more ways to share data, share infrastructure, share tools, and share learnings. There’s no time for agencies to reinvent the wheel on their own and no time to repeat the mistakes of others instead of learning from them.
It’s also a great time to submit AI related funding proposals to the TMF. Right now, they’ve got money to fund teams and proposals looking for ways to make an outsized impact with AI. So take advantage of that.
For vendors, we’re counting on you to keep innovating and building new platforms and services that align with our shared values. We’re also eager for you to engage with our FedRAMP team, understand the new roadmap and keep helping us improve the speed and efficiency of that important program. Getting that right will take your ideas and collaboration, and it’s the only way we’ll be able to effectively deliver for the people we serve.
As I close, I’d encourage you all to take a minute to look around this room, because these are people from all across the government charged with getting this right. The folks who will ensure that we stay on the right side of history as these new technology keep flowing fast. The public servants dedicated to providing easy to use services for the American people that also reflect the values of security, accessibility, and accountability.
And folks, when we do that, the impact will be profound, unlocking technology’s potential to improve people’s lives, driving innovation, growing our economy and protecting our national security.
Without a doubt there are plenty of uncharted waters ahead. The team at GSA stands ready to be your trusted guides. To collaborate with private industry. To help agency partners navigate through the uncertainty and, through it all, stay focused on showing that our government and our democracy can deliver for the people we serve.
They’re counting on us to get it right. There’s no time to waste. So, let’s get to work.