TL;DR

US state-level AI legislation is surging ahead, driven by urgency and local needs, while federal efforts stall in endless debate. This patchwork of regulations creates compliance headaches but also fosters innovation tailored to specific challenges. The conventional view that federal inaction is a total loss misses the nuance: state laws are both a stopgap and a testing ground, offering valuable lessons even as they complicate the landscape for businesses.

The setup

Artificial intelligence is transforming industries at breakneck speed, from healthcare to transportation, finance to education. As of 2023, the US has no comprehensive federal AI legislation. Instead, a patchwork of state laws is emerging, driven by concerns over bias, privacy, and the impact on local economies. This state-level activity is happening against a backdrop of growing awareness of AI’s risks: biased algorithms in hiring, facial recognition abuses, and the potential for AI to exacerbate inequality.

The conventional wisdom is that federal inaction on AI is a critical failure. Commentators lament the lack of a unified regulatory framework, arguing it leaves the US vulnerable to competition from China and the EU, which are moving faster on AI regulation. However, this view overlooks the critical role state laws are playing in shaping AI’s future in the US.

The argument

State laws are filling a critical gap, even if imperfectly

The absence of federal AI legislation has created a vacuum that states are rushing to fill. California, New York, and Illinois are leading the charge with laws targeting specific AI applications. For example:

  • California’s Automated Decision Systems Accountability Act requires companies to audit AI systems for bias and provide transparency reports. This law directly addresses concerns about algorithmic discrimination in hiring, lending, and other high-stakes decisions.
  • New York’s Biometric Information Privacy Act imposes strict consent and data protection requirements on companies using facial recognition and other biometric technologies. This law responds to growing fears about the misuse of facial recognition in public spaces and its potential for abuse by law enforcement.
  • Illinois’ Artificial Intelligence Video Interview Act mandates that employers inform candidates when AI is used in video interviews and obtain consent. This law reflects concerns about AI’s impact on employment and the potential for bias in automated hiring systems.

These laws, while imperfect, represent a crucial first step. They address immediate concerns and provide a framework for accountability that is currently absent at the federal level. However, the lack of uniformity across states creates compliance challenges for businesses operating nationwide.

State laws foster innovation by allowing for experimentation

One of the often-overlooked benefits of state-level AI regulation is the opportunity for experimentation. Different states are trying different approaches, which can lead to a more nuanced understanding of what works and what doesn’t. For example:

  • Washington State’s approach focuses on creating a regulatory sandbox for AI companies. This allows startups to test new AI technologies under relaxed regulatory scrutiny, fostering innovation while still providing some oversight. This model contrasts with California’s more prescriptive approach and could offer insights into balancing innovation and regulation.
  • Texas’ focus on AI in education is another example of state-level innovation. The state is exploring using AI to personalize learning experiences for students, with a focus on improving educational outcomes. This targeted approach allows for experimentation with AI in a specific sector, potentially leading to breakthroughs that could be applied more broadly.

By allowing states to act as laboratories for AI policy, the US can develop a more refined understanding of how to regulate AI effectively. This approach contrasts with the more top-down, one-size-fits-all approach being pursued in the EU, which some argue could stifle innovation.

State laws can address local concerns more effectively

AI’s impact varies significantly across different regions and communities. State laws can be more responsive to local needs and concerns than a single federal law. For example:

  • Rural states like Montana and Wyoming are grappling with the impact of AI on agriculture. These states are exploring regulations that address the use of AI in farming, such as precision agriculture technologies that rely on AI to optimize crop yields. The concerns in these states differ from those in urban areas, where AI’s impact on transportation and housing is a more pressing issue.
  • States with large immigrant populations, like Texas and Florida, are more focused on the use of AI in immigration enforcement. These states are considering laws that limit the use of AI in immigration decisions, reflecting concerns about bias and fairness in AI systems used by government agencies.

By allowing states to address these local concerns, the US can develop a more comprehensive approach to AI regulation that reflects the diversity of its population and economy.

What the conventional view gets wrong

The conventional view that federal inaction on AI is a total loss misses several key points:

1. The false dichotomy between state and federal action

Critics argue that state-level regulation creates a fragmented landscape that is confusing for businesses and hinders the development of a cohesive national strategy. While there is some truth to this, it overlooks the fact that state laws can complement federal efforts rather than compete with them. For example, states can act as testing grounds for regulations that could eventually be adopted at the federal level. The success or failure of state-level experiments can inform the development of more effective federal policies.

2. The potential for a race to the bottom

Some worry that state-level regulation will lead to a race to the bottom, with states competing to attract AI companies by offering lax regulatory environments. While this is a legitimate concern, it is not inevitable. States can and do regulate industries without engaging in a race to the bottom. For example, California’s strict environmental regulations have not prevented it from being a hub for technology and innovation.

3. The assumption that federal action would be better

The assumption that federal AI legislation would be inherently superior to state-level laws is not necessarily true. Federal legislation can be slow, cumbersome, and influenced by lobbying efforts that may not reflect the public interest. State laws, while imperfect, can be more agile and responsive to emerging issues.

The implications

For builders: A more complex regulatory environment

For AI companies, the patchwork of state laws creates a more complex regulatory environment. Companies must navigate different requirements across states, which can be costly and time-consuming. However, this also presents an opportunity. Companies that can adapt to this fragmented landscape and build compliance into their development processes will be better positioned to succeed.

For users: Greater protection in some areas, uncertainty in others

Users of AI technologies may benefit from the increased protections offered by state laws. For example, the Illinois law on AI in hiring could help reduce bias in automated hiring systems. However, the lack of uniformity across states can create uncertainty for users, particularly when it comes to privacy and data protection.

For regulators: A call for coordination

The rise of state-level AI laws underscores the need for greater coordination between states and the federal government. While states can lead the way in addressing specific issues, a coherent national strategy is still necessary to ensure consistency and avoid conflicting regulations.

For the industry: A need for self-regulation

The current regulatory landscape also highlights the importance of self-regulation for the AI industry. Companies that proactively address ethical concerns and implement robust governance frameworks will be better positioned to build trust with consumers and regulators.

What I’m watching next

1. The development of a federal AI framework

The question of whether the US will develop a comprehensive federal AI framework remains open. I am watching for signs of progress in Congress, such as the introduction of new legislation or the establishment of a dedicated AI regulatory body. The outcome of the 2024 presidential election could also be a key factor in determining the future of AI regulation in the US.

2. The impact of state-level experiments

I am also closely monitoring the impact of state-level AI laws. Which approaches are most effective? What unintended consequences arise? The answers to these questions will be crucial in shaping future regulatory efforts.

3. The role of international cooperation

Finally, I am watching the role of international cooperation in AI regulation. As AI technologies become more global, the need for harmonized standards and regulations becomes more pressing. The US will need to engage with other countries to ensure that its approach to AI regulation is aligned with international norms.

The bottom line

State-level AI laws are filling a critical gap in the absence of federal action. While they create challenges for businesses, they also offer opportunities for innovation and experimentation. The US can leverage the strengths of state-level regulation while working towards a more cohesive national strategy. The key is to find the right balance between fostering innovation and protecting consumers, between state and federal action, and between regulation and self-regulation. The path forward is not straightforward, but the lessons learned from state-level experiments will be invaluable in shaping the future of AI in the US.

In the end, the surge in state-level AI laws is not just a response to federal inaction; it is a testament to the complexity of regulating AI. As the technology continues to evolve, so too must our approach to regulation. The US has an opportunity to lead the way, not by imposing a single, rigid framework, but by embracing a more flexible, adaptive approach that reflects the diversity of its states and the rapid pace of technological change.