Technology is advancing every day. We believe oversight should, too.

Taka Ariga, Chief Data Scientist and Director of Innovation Lab, Government Accountability Office

As a technologist, the fascinating opportunities offered by the rapidly evolving technology landscape of web 3.0 makes me giddy.. At this year’s Consumer Electronics Show (CES), I saw how use cases for digital twins, extended reality, and various other integrated cyber-physical capabilities—all built on the foundations of artificial intelligence and cloud computing—can spark another reimagination for how we work and collaborate.

It goes without saying that the federal government is not Silicon Valley. However, the federal government is also not insulated from impacts of technological evolution, but also must help to deal with its growing pains. Nearly every federal CIO will tell you that the pace of technology modernization—kicked into a higher gear by necessity during the COVID-19 pandemic—shows no sign of slowing down. This trend towards digital-everything not only raises the level of technical complexities by orders of magnitude but also brings with it a host of challenges related to acquisition, privacy, workforce, governance, compliance, cybersecurity, and potentials for disparate impacts on segments of society. The most basic question we need to ask is: how do we know that these interconnected technologies are performing in ways that we (humans) intended?

U.S. Supreme Court Justice Louis Brandeis once said that sunlight is the best disinfectant. The intertwined challenges associated with often opaque technologies can borrow from that same sentiment, and are best unpacked through robust, independent oversight. GAO has a long track record of leading this mandate. Our north star is to objectively and consistently articulate programmatic effectiveness, efficiency, economy, ethics, and equity. GAO is now taking that legacy several steps forward by developing hands-on practitioner experience with a variety of emerging technologies through our Innovation Lab. Exploring and experimenting with technologies has become a necessary foresight function that helps GAO adapt to the emerging consequences of web 3.0 and beyond.

GAO established the Innovation Lab just prior to the agency’s centennial to explore the inner workings of machine learning techniques, cloud services, zero trust controls, blockchains, edge devices, and complex cloud-connected instrumentations. At the same time, we are also tasked to figure out how to appropriately apply these emerging capabilities in ways that enhance GAO’s own oversight capacity. It is a challenging dual mission because there is no established playbook that tells us how to handle each type of emerging technology to serve oversight functions. Yet, the bar is necessarily set very high because of GAO’s unwavering commitment to criteria and principles set out under the bedrock of our Yellow Book auditing standards. This is why it is critical for the Lab’s data scientists and technologists to develop our own muscle memory.

The journey thus far, while difficult, has been exhilarating. We started out as a startup entity and have evolved our innovation lifecycle management processes to balance the entrepreneurial spirit with critical compliance functions, all while maintaining our own brand of agility. As the Lab navigates the uncertainties across our portfolio of projects, we are making sure that complex technical work and associated risks do not jeopardize GAO’s principles of quality, non-partisanship, and objectivity.

Along the way, the Innovation Lab initiated a three-pronged strategy to orient our data science, data governance, and data literacy capacity toward the use of AI. GAO was able to publish a first-of-its-kind AI Accountability Framework that ties aspirations down to a set of implementable practices and evaluable procedures across the AI lifecycle. We developed a publicly accessible interactive policy simulation tool to better convey nuanced tradeoff decisions in order to reduce improper payments through use of identity verification controls, and tackled acquisition challenges related to the procurement of cloud services. We also developed our own topic modeling algorithms, implemented the first cross-agency blockchain prototype, and shortened our Authority-to-Operate (ATO) cycles down to days instead of months.

Today, the physical Lab space—built entirely with multidisciplinary collaboration in mind—is buzzing with activities. We are working on the use of extended reality to enhance evidence collection in insecure operating environments, developing precedent setting machine learning techniques appropriate for audit contexts, and taking advantage of open-source information to establish data pipelines that integrate automated data reliability assessment.

As for tomorrow, we will continue to grow our capacity and to build on our successes with a continued sense of urgency. We recognize that an entity like the Innovation Lab is unique within the legislative branch and we want to make sure we also have greater capacity a serve Congress in the form of technical assistance. At the same time, we are looking to move the needle on systemic opportunities such as continuous auditing, portfolio knowledge management, improper payments, and diving into the deep ends of cybersecurity implications through the use of emerging technologies. One of my assistant Lab directors is fond of quote by the French statesman Charles Alexandre de Calonne, and his words are certainly relevant here: “the difficult is that which takes a little time, the impossible is what takes a little longer.”