In March 2021, Good Systems hosted a first-of-its-kind event with the World Economic Forum, inviting researchers and industry leaders to talk about government procurement of artificial intelligence, or — simply — how state, local, and federal government agencies buy AI technologies like computer vision, used for everything from autonomous vehicles and facial recognition, and natural language processing for things like virtual assistants or chatbots. Because we live in a world where artificial intelligence is rapidly finding its way into all facets of our lives, it’s critical that we consider the long-term ramifications and ethical implications of those technologies.
But governments don’t create a lot of their own AI technologies, so they have to turn to the marketplace instead. Agencies do business with small-to-medium-sized firms as well as corporate giants like Google and Facebook to stay essential in today’s fast-moving world — and also improve the way they provide goods and services to enhance the public good. As the World Economic Forum’s Head of AI and Machine Learning Kay Firth-Butterfield says, “Procurement is important because it’s a way of influencing the ‘good’ use of AI without changing laws or resorting to regulation, which takes a long time to adopt and is difficult to enforce or amend.”
Unfortunately, though, procurement offices don’t always understand the technology they are trying to purchase. Enter the World Economic Forum, which has been a leader in developing guidelines to facilitate these kinds of processes by putting guardrails in place to assist with the adoption and implementation of AI technologies in the place of timely and effective legislation. The UK was the first to adopt them last year, and they’ve been tested by the United Arab Emirates. While the Forum’s early work has focused on national government procurement processes, our conference explored how these strategies might be applied to state and local governments as well.
The expert panelists who participated at our Week with the World Economic Forum had a lot to say about these and other topics. We have accumulated a few of the most memorable moments here. You can also view videos of each of the panels on our YouTube page.
“Governments buy a lot of stuff. Making sure the stuff they buy is ethically manufactured ensures a better AI landscape for all.”
UT President Jay Hartzell kicked off our Week with the World Economic Forum with some eloquent remarks about government procurement of AI. We know, it seems like a niche topic. However, when state, local and national governments decide to buy artificial intelligence, the choices they make and the guidelines they lay out influence the ways companies design products and employ AI technologies we all use every day. In other words, Hartzell said, government procurement is a measure of how democratic values are expressed in market settings. “Not only can such guidelines help ensure governments utilize AI in ethical and democratic ways, they also help more widely infuse those values across the market.”
Tune into Day 1 of our panels to hear Hartzell’s full remarks, including two very moving stories illustrating both the promise and perils of AI near the 4-minute mark.
“The thing that all of our secretaries of defense have said that keeps them up at night is not China or Russia or any of these other countries. In fact, what they fear the most is that, as a nation, have we lost the ability to go fast.”
Seventy years ago, the U.S. was in a bitter race against other countries to get to space. Today, we are in a similar race to adopt commercial technology for everything from defense missile systems to cybersecurity. And we’re falling behind, says Alex Goldberg, Texas Strategic Engagement Lead for the Department of Defense. “Back in the 50s, everything from your cell phone to your mouse to the internet to LCD screens was primarily developed out of the government,” he says. Today, that’s not the case. “Today, for every $1 we have, it’s the Googles, the Apples, the Facebooks, the private equity venture community that has $11. And when we’re talking about artificial intelligence, it’s $100 to every $1 we spend in investments.”
In other words, the critical technology that governments need is being created in places like Austin and Silicon Valley. Goldberg says this is why it’s especially important to remove some of the red tape to make it easier for governments to work with industry. This will allow us to use artificial intelligence for the betterment of our country so we don’t fall behind, he says. He calls it “moving at the speed of relevance” — procurement guidelines that can easily be implemented will help the U.S. stay in the game. Goldberg cautions, though, that we must also move “at the speed of trust” so that governments can be certain they can rely on AI when making critical decisions. Guidelines help to ensure that happens.
“Artificial intelligence is mainly fueled by… startups and small companies. The startups are constrained by available resources and funding. We are not Google and Facebook. If we had to spend, say, 40 to 50% of our budget dealing with all this regulation, there’s no way we can move this field forward.”
On the third day of panels, Shaoshan Liu, CEO of the autonomous vehicle company Perceptin, shared his experience trying to meet government regulations to get his product — which is often used for transportation services — to market. He described a mixed bag of requirements depending on whether the vehicles were being reviewed in the U.S. or Japan. “In the past 20 years, we have enjoyed the benefits of globalization,” he says. “But today’s government procurement process is anti-globalization, meaning that you have to deploy a very ad-hoc process against a very standardized process for AI procurement.”
In order for the government to buy his technology, the company has to get approvals for everything from privacy to security for each vehicle component. This means as many as 30 parts from countries spanning China to South Korea. Because governments typically don’t understand the technology, they don’t even know relevant questions to ask before making purchasing decisions, Liu says.
“For a startup like us, it’s very hard to do all kinds of certification and to pass all the scrutinization in order to win the contract,” he says. “It would be very nice if we had a global standard to guide each and every single government procurement project.”
“I think that gap between what we hope would happen and what actually happens when we deploy a technology can be quite big.”
Would you support governments collecting your private information if it were used to protect the public good? That’s the question panelists were asked on the last day of our Week with the World Economic Forum. And opinions varied wildly.
“People need to have a stronger mindset of how their data is being collected, what it is being used for, and what rights over the data we should have,” said Tuan Nguyen, deputy director of community and brand for SGInnovate in Singapore, who supported governments collecting personal information, especially in nations that put a lot of trust in their governments like Singapore.
Maria De-Arteaga, an assistant professor in the Department of Information, Risk and Operation Management at the McCombs School of Business, took a firm stance on the other side. She said when governments collect data, they may intend to do well but that doesn’t mean they always will. “I think that gap between what we hope would happen and what actually happens when we deploy a technology can be quite big,” she said.
Hwanyong Kim, an associate professor in architecture at Hanyang University, came out somewhere in the middle, saying that personal data collection would be beneficial to creating safer communities. But he cautioned that a “safer community” means different things to different people. What protects one person might harm someone else. “If we can provide enough understanding and get some consensus from those individuals that are going to use those data sets for something with good reason, and if those people can actually agree about the use of those datasets to make a safer community, then it could be probable perhaps.”
Tune in at the 50-minute mark to hear the discussion.
“I am grateful to UT and Good Systems because they support student participation. They collaborate with researchers from different disciplines so they can contribute in different ways. The outcome is greater and has more impact thanks to these interactions.”
One of our Week with the World Economic Forum events occurred behind closed doors (or, rather, behind a private Zoom link). This was a day reserved solely for graduate students to come together with World Economic Forum leaders to discuss the ways they could work together to help governments better employ AI. Pablo Pejlatowicz, who is earning his master’s in public affairs at UT, was one of about 20 who participated. He said the topic piqued his interest and fit with his background, having served as former director of industrial policy application at the Ministry of Productive Development of Argentina.
After exploring different ethical considerations when buying AI technologies, Pejlatowic signed up to be one of several students who will work with the World Economic Forum starting next month to roll out a set of government AI procurement guidelines for Latin America and the Caribbean. The guidelines the World Economic Forum designed for the UK may not work for Latin American countries because of different needs and systems of government, he says. “It could be the opportunity for the Latin American partners to sit down and reflect on how these guidelines could be better adapted to their own situation,” Pejlatowic said of the project. “Given my expertise in Argentina and teaching human rights at Universidad de Buenos Aires, I could be a bridge for other Latin American countries and provide support from the side of UT in this initiative.”
To watch all four panels, check out our YouTube playlist:
Please join us on this journey.
Good Systems is a research grand challenge at The University of Texas at Austin. We’re a team of information and computer scientists, robotics experts, engineers, humanists and philosophers, policy and communication scholars, architects, and designers. Our goal over the next eight years is to design AI technologies that benefit society. Follow us on Twitter, join us at our events, and come back to our blog for updates.