Government AI Hire, Use, Buy (HUB) Roundtable Series – Roundtable 1: Government as a User of AI

Table of Contents Introduction...............................................................................................................................................2Overview of Discussion.........................................................................................................................3Session I: Foundations, Ethical Frameworks, and Oversight....................................................4Session II: Security and Transparency..............................................................................................5Session III: AI Applications and Challenges, Including Bias and Workforce Issues..........7Session IV: AI in the Policy and Decision-Making Process........................................................9Authors andAcknowledgments.....................................................................................................11Endnotes.................................................................................................................................................12 Introduction The U.S. government has embraced the potential ofartificial intelligence, but there arenot yet sufficiently comprehensive strategies and guardrails, nor robustinteragencycoordination,across federal agencies to maximize AI public good applications whilemitigating risks. In light of this gap, Georgetown’sGovernment AI Hire, Use, Buy (HUB)Roundtable Seriesbrings together leading voices to grapple with the legal liabilityquestions that AI poses, examine AI’s potential to transform government services, andconsider how the government can better attract and use AI talent. The federal government has been particularly active on AI over the last two years.Following the release of several advanced large language models, President Bidensigned the groundbreaking Executive Order(EO)on the Safe, Secure, and TrustworthyDevelopment and Use of Artificial Intelligence in October 2023.1In March 2024, theOffice of Management and Budget released its government-wide policy, AdvancingGovernance, Innovation, and Risk Management for Agency Use of ArtificialIntelligence, which complements the recently finalizedEuropean Union AI Act.2Additionally, the Government Accountability Office conducted its audit of current AIuses by federal agencies.3However, many questions on the U.S. government’s use,procurement, and hiring of AI tools and talent remain unaddressed. Georgetown University’sCenter for SecurityandEmerging TechnologyandBeeckCenter for Social Impact and Innovation, together with theGeorgetown Law Institutefor Technology Law and Policy(Tech Institute),leda series of invite-only roundtablesover the course of 2024to discuss these pressing questions.This resulting report wasauthored in 2024 after those discussions. This report summarizes and reflects on the key findings, insights, andrecommendations from the first roundtable of the HUB Series: “Government as a Userof AI.” This roundtable was led by the Tech Institute and focused on the U.S.government’s use of AI technologies. The HUB Series operates undertheChathamHouse Rule;therefore,all names, affiliations, and other identifying information havebeen omitted from this report. The HUB Series is funded by a grant from the Rockefeller Foundation. Overview of Discussion The daylong roundtable featured four sessions examining the government’s use of AI: I.Foundations, ethical frameworks, and oversightII.Security and transparencyIII.AI applications and challenges, including bias and workforce issuesIV.AI in the policy and decision-making process Attendees included representatives from federal government, civil society groupsworking on AI governance issues,legislators,and industry. We provide a summary of the discussionfromeachof the sessionsbelow.*Keythemes, concerns, and recommendations to emerge from the group discussionincluded: 1.The challenge of definingAIin a way that capturesall potential use cases,benefits, and harmswithout regulating so broadly thatthe definitionincludesold technology not previously considered AI.2.The importance of developing a combination of standards, guidance, bindingpolicies, and legal frameworks to govern the use of AI, rather than relying onany one of these tools alone.3.The need to maintain diverse, human-in-the-loop perspectives to oversee AI usein government and ensure protections for marginalized groups, federalemployees, and others.4.An agreement that AI is best poised to positively impact government services indata analysis, process improvement, and other complementary roles, rather thanas a replacement for government workers.5.A sentiment that more resourcing is needed for:a.Agency-wide and individual-level AI training and education.b.Development of standards and guidance to support responsible andsecure AI adoption and integration. SessionI: Foundations, Ethical Frameworks, and Oversight ●Do we have, or can we reach, a shared understanding of what AI is and is notfor purposes of thisdiscussio