The Center for Data Innovation regarding the AI report card.

Best listening experience is on Chrome, Firefox or Safari. Subscribe to Federal Drive’s daily audio interviews on Apple Podcasts or PodcastOne.

From federal policymakers to agency implementers, we’ve heard a ton of talk about artificial intelligence (AI). But neither group has done nearly everything it said it wants to do to promote effective use of AI. That’s one takeaway from a new report card from the Information Technology Innovation Foundation. Hodan Omaar is a policy analyst who helped develop that report card for ITIF’s center for data innovation. She talked about some of the findings with Federal News Network’s Jared Serbu on the Federal Drive with Tom Temin.

Interview transcript:

Hodan Omaar: One of the things I found working in AI policy here in the United States is whenever you talk about AI policy, people are often speaking about very different things. So you know, some people could be talking about this new initiative to ensure that individuals in AI researcher have access to the compute they need. Many other people are talking about regulation, and if it’s right or wrong, and whether it should be closer to the EU or further away from the EU. Some people are talking about AI r&d. And so, the whole idea of the report was kind of to take a comprehensive view of U.S. AI policy at a high level for the United States. And really think about what U.S. AI policy means. I kind of chose the nine most prominent policies that the United States uses to kind of fuel AI innovation and competitiveness. And this report is really measuring the United States against itself. So if I say it’s approaching expectations, or it’s meeting expectations, it’s sort of the way that a report card would be for an individual child, you know, what is the potential for this individual? What is there a potential for in the United States? And how is it faring against what it could be doing? Rather than kind of comparing it against other countries, we’ve done other reports about the bottom line when it comes to the U.S. and how it’s doing in AI innovation and competitiveness looking at other countries saying, you know, the United States in general, when it comes to U.S. AI policy, we have a report out looking against the EU and China. And that report kind of says that the US is coming in first. This report, because it’s looking against the United States, its own potential against itself, the bottom line is kind of hard to say just because it varies so differently in those nine areas.

        Join us May 6 at 1 p.m. EST for Federal News Network's Industry Exchange Data where we'll explore how you can employ data to modernize your agency. | Register today!

Jared Serbu: Fair enough. I want to drill into just a couple of them. Again, for the sake of time, I want to point our listeners to our website, where we will post the full report because you do have recommendations in each of the nine areas. But just to focus on workforce as a topic. For starters, here, it’s the one area where you say the US is failing expectations. And it’s also the one area where you spend out of the nine, the most detail in the report. Talk us through what you see as some of the biggest struggles here. Sounds like a lot of it is really just sheer quantity.

Hodan Omaar: Yeah. So when it comes to the AI workforce and strengthening the AI workforce, in the report, I kind of break this section up into looking at AI education. So that’s primary and secondary AI education, also higher AI education, and then looking at things like workforce training, and then also our immigration policies. Because, you know, for the United States, it’s not about just training up domestic talent, it’s also about whether we’re able to attract and retain talent from other countries. Particularly because for AI, AI innovation is really reliant on a lot of foreign talent. What I find in the report is when it comes to AI education, it’s really quite patchy at the kind of primary and secondary level. So you know, the U.S. in general, its education approach has always been decentralized. And that’s good on one hand, because it’s good for creativity and innovation. You know, one school wants to do something, it has the ability to be able to test those things. But if that’s going well, how do we ensure that we spread that across the country, and what the report kind of finds is that there are hotspots of areas or schools or regions that have done the investments that they need to do have innovated when it comes to AI education, and their children are doing well, but the rest of the country isn’t faring as well. And how do we really address that? There are recommendations in the report for that. And then when it comes to higher AI education, the United States has always been really, really good about having strong AI programs at the kind of undergraduate and graduate level. But what the report finds is that higher educational institutions aren’t able to really respond to many of the market signals. More and more companies and organizations want people with graduate degrees in AI. And that means that students want to study it more and more. The demand for AI education at the higher level is really, really high. And U.S. institutions just aren’t able to deliver, or to provide the supply for the demand that’s being experienced. And there’s this kind of question of where’s the responsibility on institutions, and where’s the responsibility for government? And there are kind of recommendations on both ends of that because institutions have a really important role to play and being able to educate AI talent is good for the university itself. And we’ve seen things like I have another report out that came out two weeks ago, which is looking at a partnership between the University of Florida and Nvidia. It’s essentially become the first AI university looking at creating AI curriculum across the campus. And that was really a public-private partnership. And that’s the institution really doing its own thing and the state government has been involved but you know, how do we replicate that? That report really looks like that, but there’s also this idea in the report about AI credentials. And this is something that government can get involved in. Because if a student wants to have AI credentials, they can do so through their own educational institutions, like the University of Florida is created the AI credentials. But there are lots of private companies that provide these AI credentials. But you know, what is the demand for them? Are organizations really using these AI credentials, when they look to hire individuals? Do they look for AI credentials? Do they accept AI credentials, there’s no point in really creating all these AI credentials, if organizations aren’t going to use them or aren’t going to accept them. You know, if the government, which also needs AI workers, begins to accept these for itself, that could really spur private organizations doing the same thing.

Jared Serbu: A lot of those training and education initiatives that you just talked about seems like they’re going to take some time to grow the workforce that we need. It’s interesting to me that the two specific recommendations that you have at the end of the section are directed toward Congress are both related to immigration. And I wonder if that’s because you think that’s a shorter term, way to start solving this problem? Is that the right way to think about this? Short term for immigration and longer term for education?

Hodan Omaar: I would say you really need both right now. It’s just more about that you need both types. And it’s the importance it’s really more about the importance of attracting foreign talent. Foreign-born talent has always played a vital role in innovation and competitiveness. And it particularly plays an important role in AI. I think there’s an interesting statistic I point to in the report, that it’s important not just for people who are going into established companies, but really for AI startups. So I point to some other research that was done by another think tank that finds that if you look at the top 50 AI startups, you know, what Forbes describes as the top 50 AI startups. 66% of those have a founder who is foreign-born, who came here on an immigration status. And also the statistics in you know, my report about just how much the United States relies on foreign talent. It does so predominantly for AI innovation and competitiveness. And that’s why there are specific reports about updating the kind of immigration process for these individuals to make sure that they can kind of work here and want to and also looking, this is an area where I kind of also highlight what other countries have been doing. Because several countries in the last five years have updated their immigration policies to be able to kind of attract talent. They understand that this is something that’s very, very important, and they’ve updated their policies to do so. Whereas immigration policy in the United States has been stagnant for decades.

Jared Serbu: I want to hit at least one of the other policy areas in the short time that we have with you, which is promoting government adoption of AI? I think some of your findings in this area really mirror our experience covering these issues for Federal News Network, which is enormous amount of talk about AI and government with relatively little adoption. Is that a fair way to put it and then maybe talk a bit about why you think it’s important for government to be a leader here?

Hodan Omaar: Yeah, I think that is a fair way to put it. Again, I think when government does things, it really spurs adoption in the private sector. So there’s a kind of twofold role here for government, it’s one using AI to kind of complete its own missions, because AI will help in many government missions across agencies. But also to signal to the private sector the benefit of doing so and kind of promote best practices to ensure that they do so in ways that are kind of good, aligning their operating budgets to AI innovation. Another issue is just not having direction from the top. A recommendation in the report is that every federal agency should develop its own AI strategy. We’ve seen this from you know, the Department of Health and Human Services have got one, you know, the FDA have been doing work in this, the Department of Veteran Affairs, and every agency should really think about what is our direction? What are we trying to do? Because if you don’t have direction from the top, how are you going to really spur adoption? Another thing is procurement. Some of the research in the report looks at how, when you look at the kind of federal contracts, where they go to it’s really concentrated around the East Coast. Like I think it’s about I say 87% of federal contracts awarded for robotic process automation went to companies in Virginia and New York. Being able to ensure that across the country, the best firms are able to contract with the government, that’s going to really help to ensure that the government has the best products for the best price. And one way to do that is to kind of develop an all-encompassing procurement process, whether that be through a kind of an all-encompassing procurement website for federal AI contracts. And that could be done through the AI center of excellence, which is now situated within the Government Services Agency. So that’s another report. And finally, one of my final reports in this area is really about calling out agencies that aren’t innovating with AI. So the Government Accountability Office and the council of the inspectors general shouldn’t just be looking at waste, fraud and abuse, but they should really be looking at the waste and inertia from a lack of AI innovation and calling out agencies that aren’t doing enough to adopt AI.

Jared Serbu: Just want to spend one more beat before we have to let you go on that recommendation that you have to GSA. Because I think it’s really important and it also resonates with I think what we’ve seen in our coverage, which is so many little pilots around AI in little pockets of the federal government, and no one’s really tried to do anything big. You talk about core processes, you’re talking about what you mean.

Hodan Omaar: When I was thinking about this recommendation, is really something similar to what we’ve seen at the Department of Defense, where they have a centralized kind of body and agency that is looking at collecting kind of identifying core processes within DOD that could be transformed with AI, and making sure that every different part of DOD is able to hone in and be able to access that. So you don’t have this replication and also kind of waste from different parts of the agency, not being able to know what’s going on, and then having to do it themselves and just kind of wasting time and money. So the recommendation here is for the AI center of excellence situated within GSA to identify kind of 20 to 50 core processes that could be transformed with AI, things like customer service, or chatbots. Chatbots are a really great way to use AI to improve the way that the federal government deals with citizens and consumers. And it’s also one of the best ways to innovate with AI. Because one of the things that we often talk about is how do you build trust around AI. And if a government agency is using AI in a way that’s going to be visibly helpful to consumers, it’s going to help with trust, because let’s say, you know, I’m going down the road and the Department of Transportation has used AI to improve the road. I might not necessarily see that, you know, and so I don’t know that the reason that I don’t have potholes is because you’ve used AI to somehow identify where there are potholes and send people to fix that. But if the Department of Transportation is using an AI chatbot, that’s really, really good. And I’m trying to get bus times or something. And I know that I’m using AI or the Department of Transportation is using AI in a way that’s useful for me, it not only builds trust, but it’s builds my support as a kind of resident or your support as a citizen for the government using AI. And so what are some core processes that would not only very easily transform a government agency, but also improve the trust and the ability for the agency to continue to innovate.

        Read more: Artificial Intelligence

Tom Temin: Hodan Omaar is a policy analyst at the Center for Data Innovation.

Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.