Preparing the Texas workforce for artificial intelligence

A version of the following testimony was presented to the Texas Senate Committee on Economic Development on September 22, 2026.

Artificial intelligence is already changing the way some workers do their jobs. While the available data do not yet show evidence of job loss directly caused by AI, larger disruptions in the coming years are likely. Like other recent technologies that changed the way people work, such as personal computers and the internet, these changes will be the result of a learning process by firms and workers and are difficult to predict in advance.

This gradual and less predictable trajectory is good news for workers. No substitute exists for the on-the-job learning that information economy workers have already been doing. Even the best-designed training programs or policy interventions cannot replace the skills that workers will gain by using AI in their jobs now.

This is why policymakers should avoid inflexible mandates about how firms adopt AI. For example, legislation that attempts to “protect” workers by restricting workplace AI adoption would actually jeopardize their future in two ways. First, such firms will likely become less competitive with time, ultimately having no choice but to lay off workers. Second, the individual workers will find themselves without the new skills that will be in demand.

Texas, with its free-market culture and emphasis on private-sector-led growth, is therefore well positioned to let this process of “natural reskilling” take place.

But while competition and individual worker adoption of AI are the most important ways to address any coming displacement, certain groups—such as older workers and those in fields where adoption might naturally be delayed could find themselves left behind, and there could be smaller numbers of individuals in all professions. This is where education, training, and reskilling programs come in.

State funding of job retraining programs must be approached with caution. Examples abound of government-funded retraining programs motivated by major economic events—such as the Great Recession and offshoring of manufacturing jobs—that simply didn’t deliver results for those they were intended to help. Given the lack of current evidence of AI-driven job loss, now is not the time to implement such state funding. However, if job displacement increases, these questions will likely be raised in the future. In this case, there are best practices to follow.

Government-funded training and reskilling programs have historically required grant recipients to submit extensive plans before receiving funding. While ensuring programs are reputable and effective before receiving funding is understandable, this approach presents major problems when new technology or global trends necessitate retraining workers. For example, a 2006 federal program called Workforce Innovation in Regional Economic Development (“WIRED”) sought to retrain workers as manufacturing firms moved overseas. It distributed $325 million across dozens of regions and programs. However, recipients that met the extensive planning and paperwork required for grants could not meaningfully adjust to the greater numbers and changed needs of workers in the wake of the 2008 economic crisis. Such points of failure are likely to be even more costly under the fundamentally uncertain conditions of continued evolution and adoption of AI.

Instead of funding a small number of large, thoroughly pre-planned programs in response to AI, Texas should foster an ecosystem of many smaller, more diverse initiatives. This approach offers several advantages. First, a large number of small programs could reduce the risk of funding bad initiatives that may arise with fewer early vetting requirements—successful programs could be replicated and receive more funding, while lower-performing ones could be allowed to fail quickly. Second, programs could be scaled up or down over time as the extent of AI workforce displacement becomes clearer. Third, smaller programs with an entrepreneurial approach would be more likely to experiment and hit upon approaches that could not be easily pre-planned. Local institutions like community colleges are well-placed to learn and adjust in as close to real time as possible.

Many workers will learn new skills most effectively in their current jobs, and efforts to regulate workplace adoption of AI from the top down will likely backfire. Similarly, large pre-planned education initiatives may look good today but are almost certain to fail amid AI development that we can’t yet predict. Allowing workers to learn on the job while taking a portfolio approach to many smaller training programs with private and community partners will likely deliver the best results for Texas’s workforce in these uncertain times.