Since ChatGPT launched in 2022, generative artificial intelligence (AI) has revolutionized labor. A Mercatus poll finds that use of large language models (LLMs) at work has increased 15.8 percentage points over the past six months, with workers estimating it allows them to complete research and writing tasks in one third of the former time.
However, multiple studies find that companies are not reaping proportional advantages when compared to individual employees. McKinsey researchers say employers are struggling to capitalize on the technology by organizing company-wide implementation, and the University of Lausanne notes that although 82% of employees polled said AI saved them time on their work tasks, 36% admitted to wasting more than half of what they saved.
Lack of employee efficiency will lead to their replacement, and it also increases purchase costs and stifles technological advancements for them as consumers. Thus, enabling employers to maximize the potential of their employees’ AI use helps everyone concerned.
To that end, President Trump signed an executive order to begin the American AI Initiative in 2019, aimed at preventing restrictions on AI use while directing resources toward its adoption. It outlines plans for educational opportunities in the field, including the launch of national AI research institutes. Since returning to office, he’s signed several other AI-related executive orders designed to improve America’s AI competitiveness, especially relative to China, with special consideration given to streamlining capital expenditures to AI projects.
Congress has begun navigating the AI landscape as well, with federal efforts to prevent state regulations seeing mixed results. Although the House voted to impose a decade-long moratorium on state AI restrictions, the Senate ultimately removed this stipulation from the proposed legislation.
However, legislators can still aid in this endeavor by rescinding counterproductive, antiquated labor laws. Loosening regulatory barriers aimed at a pre-AI economy will facilitate a smooth transition to the modern work ecosystem. As Forbes notes, “The extent to which automation affects the job market will largely depend on how effectively economies can adjust to these shifts.”
Wage Models
Legislators might begin by addressing unwise wage regulations. A National Bureau of Economic Research (NBER) working paper estimates that California’s abrupt recent fast food minimum wage increase, from $16/hr. to $20/hr., eliminated 18,000 fast food jobs. This came as fast food establishments began experimenting with generative AI, leading to further concerns about employee job security.
Although theories about minimum wage effects on employment remain controversial, legislators must acknowledge that if the cost of human employees overtakes that of using technology, employers will increasingly turn to the latter. However, most customers prefer some human interaction to strictly AI customer service, so an AI “workforce” may well not be the employer’s preference. Loosening minimum wage requirements will cut employer costs and allow them to retain the desired employee/AI balance.
In addition, employers need increased flexibility and incentives to pay with tech training assistance instead of solely paychecks. Apprenticeship and internship options should proliferate in this environment, but labor laws place roadblocks in the way. The Fair Labor Standards Act (FLSA), which ostensibly exists to aid workers, causes the opposite result when it mandates employers to pay in monetary wages. For instance, training must sometimes be categorized under hours worked, leading to costly overtime pay rules for the employer.
According to the International Monetary Fund, “Policies for accessibility, diffusion, and skills to help realize the full potential of AI are currently weak … And since jobs will change with AI collaborators, retraining and new skills acquisition deserve priority attention.”
Tax Laws
In coordination with increasing legal reimbursement options, lawmakers should examine more potential tax exemptions for employers in the new economy. Unlike subsidies, tax breaks allow both employers and other taxpayers to keep their own money. Current options are complicated and often result in failure to claim legally owed deductions, but the recent One Big, Beautiful Bill Act extended some simple breaks on earnings, such as overtime and tips.
However, the laws are still overly restrictive. For example, employers can only deduct $5,250 of their educational assistance programs from federal taxes. Education at even a public state college can reach $63,000 per year, and putting a limit on deductions shortchanges the employer and disincentivizes aid that could help employees modernize their knowledge and skills.
Benefits
Employers have confidently hired temporary freelancers for years because they did not come with the “regulatory baggage” of traditional full-time employees, and this has made constant rehiring a reasonable tradeoff. Generative AI lends itself beautifully to this gig work, but legislative attempts to mandate typical benefit packages for gig contractors may end this ideal match.
Many freelancers (including this writer) have benefited immensely from this flexible relationship, and providing alternatives to hardline benefit laws is of great importance to the nearly 12 million independent contractors across the country. For example, many states are now piloting portable benefit programs, which enable workers to keep and control their own accounts instead of having them attached to an employer. These plans provide options for incentivized employer contributions, making them mutually desirable.
Conclusion
Creative destruction is a constant challenge in a world increasingly dominated by technology. AI brings uncertainty and change, but so did the end of the horse and carriage, the handwritten book, and the slide rule. The transition away from those ultimately resulted in great gains, and generative AI will do the same. But labor policy will play a large role in determining whether that switch is pleasant or painful.