Rep. Jacobs: Washington Is Getting the AI Race All Wrong | Opinion

U.S. Representative Sara Jacobs argues that Washington's approach to the AI race with China is misguided. She contends that merely building the most powerful models first holds little strategic value if those models aren't adopted or if control over them is lost. In fact, China has demonstrated this by rapidly adopting and utilizing Chinese AI models, which have become more popular and affordable than their American counterparts. Despite leading in advanced computing infrastructure, the U.S. has fallen behind in model adoption.

Jacobs emphasizes that the key to success lies in creating AI systems that people trust and prefer using. This is where Congress has the opportunity to make a significant impact. She proposes four areas of focus: societal, model, deployer, and impact. Societal considerations include ensuring that AI incentives align with societal goals, perhaps by taxing capital on par with labor and finding ways to tax AI usage.

At the model level, voluntary self-regulation by companies has proven insufficient, as evidenced by OpenAI's recent mishap. Congress should provide regulatory oversight, including the authority to set safety standards, enforce them, and require developers to publish safety plans detailing their testing, safeguards, and contingency plans.

Deployer-level oversight is crucial, with regulations tailored to the stakes involved in various applications of AI. Federal agencies need stronger authority to enforce anti-discrimination laws when AI systems are involved, and employees who raise safety concerns should have legal protection. Finally, at the impact level, Congress must address the broader societal implications of AI, including job displacement, wealth concentration, the spread of deepfakes, and workforce preparation.

By implementing these guardrails, Jacobs argues that the U.S. can build AI systems that people trust, ensuring lasting adoption and victory in the AI race.