Since I recently held a series of congressional hearings on artificial intelligence (AI), I’ve been asking constituents, friends, and strangers what’s the first thing that comes to mind when I say the phrase “artificial intelligence.” If my interlocutor is older than me, he or she describes something close to HAL 9000, the scary computer from Stanley Kubrick’s sci-fi film 2001: A Space Odyssey. If the person is younger than me, he or she describes something like Ava, the manipulative and willing-to-kill humanoid robot played by Alicia Vikander in Alex Garland’s thriller Ex Machina. Having talked to government agencies about their engagement with AI and spoken with folks in the private sector about the challenges and opportunities AI brings, I’ve learned we are far from a HAL or Ava scenario becoming reality.
The government adoption of AI will not bring about a government being run by robots. Instead, our government will continue to be run by people, with help from algorithms dramatically improving government services for all Americans.
The government should urgently speed its adoption of AI to reduce the amount of time individuals spend unnecessarily interacting with the government and increase the speed of government response to citizens. For example, the private sector is currently using robotic process automation (RPA) to replace lost and stolen ATM cards. RPA expedites this burdensome process by automating the steps of inputting customer information to the database, cancelling the card, flagging recent charges to detect unauthorized spending, creating a requisition for a new card, inputting the new information into the database, tracking shipment, and activating the card. From reporting the card missing to having its replacement in hand, this process without RPA takes at least a week. But with RPA, the totality of these steps takes seconds, allowing the customer to get his or her card faster. So why can’t we automate government services like renewing passports?
The federal government has thousands of routine business practices that require the time and attention of many federal employees. By automating these practices through AI, the Deloitte Center for Government Insights estimates that we could save at least 96.7 million federal hours out of 4.3 billion hours worked each year, potentially saving $3.3 billion dollars annually.
Documenting and recording paperwork, managing services like passport renewals, and processing patent applications are practices that could all be dramatically improved with robotic automation. A great example of how AI has streamlined a government operation is the Army website’s use of an interactive virtual assistant that answers questions of potential recruits. The chat-bot does the work of dozens of human recruiters with a 94% accuracy rate. Imagine the cost savings and improved user experience if all government websites employed this seemingly basic AI-driven feature.
While introducing AI into the government will save money through optimizing processes, it should also be deployed to eliminate waste, fraud, and abuse. AI enables the consumption of large amounts of data, which can be analyzed for patterns, anomalies, and duplication. In 2016, the Government Accountability Office reported that overpayments in the Medicare program totaled approximately $60 billion, almost double what the National Institutes of Health spent on medical research that same year. Using AI, the government could identify these overpayments more quickly, allowing our direct investigators to focus on the costliest overpayments, recouping dollars and saving money.
Additionally, the government should invest in AI to improve the security of its citizens. Russian President Vladimir Putin said that the nation that leads in AI “will be the ruler of the world.” China is taking that mantra seriously. In July of last year, Chinese President Xi Jinping released a plan to invest heavily in AI and make China the AI world leader by 2030. It is in the interest of both our national and economic security that the United States not be left behind.
While AI will create numerous opportunities to improve our lives, it will not be free of challenges. Cyber warfare is evolving so rapidly that in the very near future, cybersecurity will be about good AI defending against bad AI. Consider the Russian disinformation campaigns of the past year: With AI, fake news will be exponentially more convincing, with the potential to create videos of people making statements that they never made, as featured on Radiolab’s Breaking News episode last summer.
We also need to protect against AI that is unfair, exclusive, insecure, harmful, and unaccountable. But accountability goes beyond auditable algorithms and data sets. It includes responsible data management to protect people’s privacy, along with ethical design to instill values in AI.
Steps are being taken to make intergovernmental AI capabilities a reality. For example, federal agencies are finally in the process of modernizing outdated IT systems thanks to a bill I authored that was signed into law in 2017. Antiquated IT systems have cost the government tens of billions of dollars annually to maintain, and now agencies are allowed to keep savings from replacing legacy products, instead of the previous use-it-or-lose-it approach that has plagued government technology for decades. Even so, actually implementing the new systems takes time.
The Emerging Technology Office within the General Services Administration has taken the lead in finding a pathway forward for agencies to implement AI by leveraging knowledge across the government, but no concrete plan of action currently exists.
And my IT Subcommittee in Congress is committed to driving the needle as well. But we must make AI implementation a higher priority nationally and find immediate solutions to these challenges.
AI is the path to a more efficient and effective government. If the Deloitte study is right, we lose out on $8.2 million in savings every single day that we wait. The future is already here and we can either lead it or be led by it.
A former undercover CIA officer, entrepreneur, and cybersecurity expert, Will Hurd is the U.S. Representative for the 23rd Congressional District of Texas. In Washington, he serves on the House Permanent Select Committee on Intelligence, as Vice Chair of the Maritime and Border Security Subcommittee on the Committee for Homeland Security, and as the Chairman of the Information Technology Subcommittee on the Oversight and Government Reform Committee.