In 2024, the Washington State Legislature created an AI Task Force. Lawmakers acknowledged that AI “holds extraordinary potential and has a myriad uses for both the public and private sectors.” It also highlighted that AI could “help solve urgent challenges, while making our world more prosperous, productive, and secure when used responsibly.” The clear emphasis on the importance of harnessing the upsides of AI reflected the fact that many Washingtonians are struggling under the status quo.
Yet, three years later, the Task Force failed to meaningfully help the state move the dial on leveraging AI to improve human flourishing. This outcome was in many ways predictable. The Task Force’s membership included few representatives from the communities that stand to benefit most from AI.
Unsurprisingly, its agenda was therefore dominated by how to confine AI use. While some of those limits may be warranted, the Task Force’s resources would have been better spent devising novel regulatory frameworks that aligned with the direction outlined by the legislature: deploying AI to solve the state’s numerous policy challenges.
The Task Force’s work must be analyzed in the context of the problems that the legislature likely had in mind when thinking about AI as a tool to improve the well-being of residents. As reported by the University of Washington, the state is among the lowest in the nation when it comes to providing people with mental health support. The stakes are particularly dire in rural communities; about half of the state’s counties do not have a single working psychiatrist.
Young Washingtonians are especially in need of greater mental healthcare; about one in five require clinical care for depression and anxiety. As discussed below, there are unique AI tools-distinct from generally available models and chatbots-that could go a long way toward addressing these and related issues.
Progress can and should be made on other fronts, too. Washington lags behind many states on critical education metrics, for instance. High schools are failing to adequately prepare many Washingtonians for success in a tight job market, let alone for meaningful participation in a deliberative democracy. For example, Makena Simonsen is suing her school district based on allegations that she was permitted to graduate despite being illiterate (notably, she purportedly earned a 3.87 GPA at Lynwood High School).
A whole set of additional problems could be outlined here. If the Task Force had been more focused on AI as a tool rather than as a problem in its own right, then I would be in a position to list all of the ways in which the state was conducting thoughtful policy experiments on how to deploy AI tools focused on those challenges. This is not to say that AI tools cannot result in harm nor to suggest that AI is the answer to every problem. Instead, this is a reflection of the fact that innovators across the country are working on AI tools intended to be leveraged in critical domains like those spelled out above.
Researchers at Duke, for instance, have developed a predictive AI tool that may help identify adolescents at risk of mental health crises months before those issues emerge. If broadly deployed, it could help the state’s overburdened mental health system by better triaging cases. Relatedly, tools such as StudyFetch have documented research that their educational AI tools are not only being responsibly used by students (i.e., not to cheat) but also result in significant improvements in retention and comprehension. Pilot programs trialing these tools could help the state augment the work of its incredible teachers and address resource gaps.
Analysis of the Task Force’s outputs suggests that it paid inadequate attention to imagining a future in which AI “goes well” and helps improve upon the status quo. Of its 11 recommendations, only two — establish a grant program in AI development and invest in K-12 STEM and Higher Education — were explicitly tied to this problem-solving approach to AI governance.
Notably, neither of those proposals resulted in new law. Instead, the Task Force boasted its contributions to new laws that likely would have passed even without study by the Task Force: placing guardrails on companion AI chatbots; reducing the use of AI in healthcare prior authorizations; requiring law enforcement to disclose certain use cases of AI; and bolstering the enforcement of child sexual abuse materials law. States around the country enacted similar legislation.
If Washington is going to lead in AI, then it must take on the difficult work of assessing how AI can improve and, yes, disrupt existing systems and institutions. Utah’s embrace of “regulatory mitigation,” a means of organizing statewide trials of AI tools while also ensuring there are adequate safeguards and reporting mechanisms in place, makes clear that such leadership is possible.
Yet, task forces made up of only one representative of “private technology industry groups” and one from the “software industry” and zero individuals tasked with representing the interests of those who stand to benefit most from AI (including young people and rural community members) are unlikely to accomplish that task.
Kevin Frazier is the Director of the AI Innovation and Law Program at the University of Texas School of Law and a Senior Fellow at the Abundance Institute.