Safe, secure, trustworthy: Unpacking the AI executive order’s impacts on CX in government

In late 2023, President Biden issued an executive order that established new standards on how the US should develop, regulate, and use AI. The Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence is significant both in size and scope, with directions for over fifty federal entities on over a hundred specific actions that guide how AI will be developed and deployed.

As AI continues to rewrite foundational operations and everyday services, it also reveals new risks and vulnerabilities. Fundamentally, this executive order will provide guidance for safe development and accountability for responsible AI use, for the benefit of all American citizens, consumers, and workers.

In addition to addressing concerns around national security, public health, privacy, and advancing American technological leadership, there is another concern this executive order addresses: rebuilding trust between citizens and their government.

The relationship between CX and trust

There is a direct relationship between how a citizen experiences their government and how much they trust their government. Many administrations have attempted to increase trust by improving experience with different government agencies. The current administration made CX a foundational priority — President Biden’s December 2021 executive order, Transforming Federal Customer Experience and Service Delivery to Rebuild Trust in Government, directly addresses the link between customer experience, service delivery, and trust.

Unfortunately, customer satisfaction with the federal government remains low. According to the Federal News Network, a recent study from the Partnership for Public Service found that only 35% of Americans trust the federal government, and 73% believe the federal government does not listen to the public.

In September, before the executive order was announced, officials from the White House Office of Science and Technology Policy, the Department of Homeland Security and Department of Defense testified before the House Subcommittee on Cybersecurity, Information Technology and Government Innovation. Speakers discussed how different government agencies were safely deploying more and more advanced AI systems to improve operations, create more advanced data analytics, and improve government customer service and outreach. A representative for DHS noted that AI systems have been used to resolve cold cases, combat cyberattacks, and even help citizens escape abusive living situations.

Craig Martell, the chief digital and artificial intelligence officer at the Department of Defense, noted that AI was an invaluable tool for supporting national defense operations and strategy: “We need different algorithms, different success criteria, and different data to train the different models underpinning each of our different use cases. It's really important to us to build systems that aren't simply dependent upon that algorithm, but that have humans wrapped around it. It's really human-machine teaming so that a human can say ‘oh, no, they got it wrong.’”

A double-edged sword

One major component of the order is provisioning specifics on oversight. And it gets very specific: Different departments are tasked with timelines and goals around creating guidance that can help ensure that the many, many uses of AI use remain compliant with federal law—the twenty-thousand-word executive order is nothing if not thorough. However, while specific instructions around guidance might be vague, outlining which agency is responsible for what oversight will hopefully create flexibility—a necessary quality when dealing with a technology that has so far outpaced regulation.

Several sections outline the responsibilities of different government agencies in how AI can be used to maximize program access for eligible recipients, when parties must be notified that AI is being used for decisions, regularly evaluating how decisions are being made, and creating consistent processes to appeal denials to a human board of reviewers. It also requires analysis post-implementation, if algorithmic systems are helping these programs “achieve equitable and just outcomes.”

Public services in the United States have an incredible scope, and any change to how those services are rendered has an incredible impact, for better or for worse. In These Times reported that “…the expansion of the child tax credit during the Covid-19 pandemic found that increasing the benefit amount and making it available to more people lifted 3.7 million children out of poverty.” AI has tremendous power to make these public benefits and services more accessible to more people, like making information available 24/7 in multiple languages and across multiple channels. It can also help on the backend with processing applications to get people faster access to the help they need, such as Georgia’s 46,000 backlog of SNAP applications. The IRS has mentioned implementing AI tools as part of an effort to chase down repeat tax dodgers.

Unchecked, AI can transform a streamlined experience to a Kafkaesque nightmare. Nobody thought that the Medicaid unwind was going to be simple, but early estimates did not account for the number of “procedural terminations” of eligible people due to a faulty automation system, according to a new lawsuit. Good CX can improve trust in government, but the inverse is also unfortunately true.

90-day progress report

According to the White House, all federal agencies have so far met the 90-day actions set by the executive order. More deadlines are coming in April. Only time will tell the full results, but there has already been an impact. One new test of this executive order? The Medicaid unwind lawsuit mentioned above.