Build US Federal Capacity to Rethink and Rein in the ‘Wild West’ of an Increasingly Data-Driven and AI-Driven World
KEY MESSAGES AND ACTIONS
Given the rapid evolution of data sharing and artificial intelligence (AI), federal lawmakers are under pressure to ensure that these critical technologies “protect the security of the American people, expand economic prosperity and opportunity, and realize and defend democratic values.” AI technologies, which rely on large and privacy-sensitive data, like Chat-GPT for intelligent chatbot queries or Stable Diffusion for generating images from brain activity, have unleashed wide, unforeseen, and still unimaginable applications in the private and public sectors. Such applications need careful review and urgent regulatory ingenuity, especially for critically vulnerable areas in health, work, education, criminal justice, finance, and international affairs that are emerging or have long stood as ‘Wild West’ zones in the digital world.
To match the scale and speed at which data and AI are changing, building federal capacity is necessary for US citizens to access high-quality digital government and for the US to establish a concerted digital strategy. Without such federal action, US citizens are at risk of being harmed by unstable digital infrastructure and the nation will lag behind in digital services and information on the global stage.
Federal policy makers should respond to these rapid technological advances by
1) creating AI trustworthiness and data traceability standards
2) ensuring compliance with principles of good AI use
3) establishing multi-stakeholder governance committees for independent review
4) negotiating for multilateral agreement(s) in favor of a free and open internet
CHALLENGES IN RESPONDING TO A DIGITAL WORLD AT SCALE AND WITH SPEED
The fast pace of an increasingly digitized, interoperable, and distributed world has created a lag in US federal capacity to prioritize how time and resources should be spent so that the use of data and AI-powered systems (Box 1) do not threaten the rights of the American public and global access to a free and open internet. Likewise, a growing data economy and the commercialization of generative AI-systems that create their own content requires regulatory ingenuity to respond to both known and unforeseen risks. Without new political imagination and federal regulatory guardrails, Americans lose out on safe and effective systems that shape how Americans work, play, and receive essential services.
To have a targeted and nonprescriptive federal response to rapid digital changes, lessons can be learned from, for example, cryptography’s regulatory history; however, the scale and speed at which data and AI are changing (Figure 1) international trade and the private and public sector is unprecedented. Nearly 5 billion people use the internet, and a single internet user can produce enough data equivalent to 150,000 high-resolution photos per day. In 2021, the digital economy contributed 10.3% of US GDP and US exports of digital services surpassed $5.3 billion. Then, last year’s introduction of a popular AI-powered system, Chat-GPT, scaled rapidly with 100 million users in the first two months, placing the chatbot with the fastest recorded growing user base. However, no bill has been proposed to curb AI’s potential dangers and 44% of Americans think they do not regularly interact with AI while 59% understand little or nothing about what companies do with the data they collect.
Gaps in federal response and in American perceptions on the digital world allow for historic or emerging ‘Wild West’ zones in the digital world (Box 2) to operate without regulatory guiderails. For example, domains that have persisted, to this day, as radically underregulated include supply chains and, cross-border data flows. Controlled by the private sector, supply chains are under no obligation and have no incentivization to disclose their data. They also have no legal and international agreements or standard fora and templates to help assess how AI, data, and emerging technologies can impact supply chain resilience and security. At best, supply chains and data flows are often considered business transactions operating under reactive rather than proactive federal support.
Other domains likes food and health have quickly and transparently assimilated AI and data-driven technologies into their daily practices. However, there is a need to expand or authorize roles that include AI so that AI tasks can be approved or cleared rather than immediately performed. These definitions and procedures can help ensure patient and consumer safety beyond simply classifying AI as “assistive” or “automated.” In total, recent and historic digital ‘Wild West’ zones can both be at risk to digital misinformation and digital authoritarianism that can undermine human rights and result in social and economic coercion.
To comprehensively approach ‘Wild West’ zones, there are several principles outlined by the White House blueprint for an AI Bill of Rights and demonstrated in proceedings from the National Academies of Sciences, Engineering, and Medicine that can be referenced (Table 1):
FEDERAL CAPACITY AS A FIRST RESPONSE TO THE DIGITAL WORLD
Prescriptive guidance on data and AI regulation are useful, but only if there is the federal capacity to address them. Recent advances for Congress to establish a National Artificial Intelligence Research Resource (NAIRR) task force — a national cyberinfrastructure similar to Amazon Web Services (AWS) — will champion AI and data standards through computing services and resources. Also, key legislation with the authority to take federal action on AI have included the CHIPS and Science Act of 2022 as well as the National AI Initiative Act of 2020 — each can allocate research funds and establish advisory commissions. Finally, Department guidance, such as from the DOJ or DOC, could help define use cases or carry out investigations on data and AI malpractice. In turn, the US has both emerging and existing entry points where high-quality digital governance can be catalyzed.
However, it is uncertain whether this federated capacity can extend beyond AI academics and be adopted by vulnerable sectors. Likewise, use cases and investigations on AI regulation may set precedence over time but slowly. To ensure that digital ‘Wild West’ zones are reined in sustainably and at scale, Congress and government agencies can build federal capacity in the following ways:
- Federate and incentivize AI trustworthiness and data traceability standards, beginning with domains most vulnerable to data-driven and AI-driven changes. The rule of law should be placed at the core of AI and data systems to ensure they comply with constitutional principles and protect human rights. In practice, this could be resolved with checklists and ongoing monitoring, third-party verification, and domain-specific review boards. Likewise, although a challenge, efforts to assess individual AI and data components should not be divorced from their entire sociotechnical systems. Officials who lack understanding on the subject matter may miss these consequential leadership opportunities.
- Ensure compliance with principles for good AI use and development with a total product life cycle (TPLC) approach. Any federated digital infrastructure should be co-located with insurance and compliance agencies for easier commercialization. Security constraints may limit these reviews but domains exhibiting security through obscurity (STO) should be converted to more productive approaches where possible.
- Establish multi-stakeholder governance committees with multiyear evidence-building plans on safe and effective digital systems. To speed their development, federal funds should be increased and expertise outsourced to a commercial entity. However, large central programs should be careful to not overshadow small, more experimental pilot programs.
- Negotiate and develop strategies for a multilateral agreement on digital trade within the World Trade Organization (WTO) in favor of a free and open internet. Diplomatic activities should increase to help prevent the abuse of AI and data systems by authoritarian regimes. It is also crucial these efforts lead quickly rather than falling behind reactive cross-border bans.
CONCLUDING REMARKS: THE NATION AND GROWING AI AND DATA DEPENDENCY
AI and data have become critical national assets and international territory to the point where they have been described as a ‘special resource’. It can very well be that AI and data systems can transform from a special resource to a life source, drastically making the way we build economies, define our cultures, conduct science, and govern increasingly dependent on the digital world. In turn, as Americans continue to connect and learn from the digital world, this work urges federal lawmakers to take the necessary next steps in democratizing and stewarding the AI and data biome.
