Do your homework on the good of AI. Then legislate it.

Inside the beltway, artificial intelligence (AI) has rocketed up the public policy agenda. While there’s policy discussion of the promise and perils of AI, the latter is emphasized. Instead, we need more policy balance with increased focus on the tremendous potential for good, and how to best ensure that this good is realized for all.

AI has great promise for our future social advancement, national security, and economic success, including how it can maintain and strengthen U.S. leadership in the global marketplace.

The good of AI is wide-ranging. Education and learning may be customized to an individual’s goals and needs, adjusting to fast-learners, slower-learners, and special interests. Students and learners won’t have to be constrained by the contours of a static curriculum or module. Aided by AI, workers throughout the economy can become more productive and provide services previously not possible or viable. New economic opportunities will be generated, by both America’s largest companies and smallest businesses.

AI should be viewed as “augmented intelligence,” not as a replacement for humans. Some tasks will be more ably handled by AI, leaving more sophisticated tasks for the human touch.

As with previous technological innovations, there will be a disruption in the workforce, with some jobs eliminated and new ones created. As this shift in employment is likely significant and will occur across most sectors of the economy, public and private investment and concerted policy efforts will be needed. The scale of the AI transition may well be comparable to the transition from the manufacturing to the service economy in the 20th century. In this previous shift, blue-collar workers faced the most upheaval. This time, it will be knowledge workers; those employed in fields with large bodies of codified information, standards and procedures.

Instructors will need training and resources to educate students of all ages on the concepts, opportunities, and challenges enabled by AI. We will need to invest in educating and training the workforce, through universities, community colleges, libraries, workforce centers, and other community organizations. Libraries, for instance, have helped the public keep up with advancements in technology by providing computer labs, audio/visual studios, and maker space labs, as well as teaching related skills and applications. The same public access and education should be broadly offered for AI-based systems and services, such as this recent workshop on ChatGPT and other AI programs at Chicago Public Library. Investments are especially important to ensure opportunity for those in low-income communities—community anchor institutions such as libraries are well-situated to provide AI-related education and awareness, by leveraging their existing physical and intellectual resources, presence and reputation.

While we can cherish the power of the free market to propel advances for the nation, leaving things entirely to the market will not be sufficient to effectively guide us through the AI transition. Public sector investments will be needed.

While emphasizing the promise of AI, there are valid challenges and concerns to consider and address—as has been true for past innovations. Questions about the appropriate use of copyrighted materials by AI have emerged, as well as the implications of this use for authors and creators. Other information policy issues are implicated such as privacy and cross-national laws. Another important issue is protection for the general public from potential AI-related harms, which include such moral and ethical concerns as advice about health, safety and other matters of personal import become yet more dependent on technology.

In a recent survey reported by Vox, 72 percent of Americans said that they want the adoption of AI to slow down. Legislation to regulate AI should also slow down, despite the energy for legislative proposals on AI. It is not yet time to act. Rather, more brainstorming and thinking are needed. We need to analyze and plan. The time is perfect to initiate a study by the National Academy of Sciences, Engineering, and Medicine—getting America’s best thinkers to develop visions, scenarios, a blueprint for strategy and potential legislative proposals in a non-partisan setting. Ramping up expertise in other ways, such as strengthening expert staffing in federal agencies is also now desirable. AI policy is too important to act in haste.

ChatGPT agrees with me. I asked, “Should the US Congress pass new legislation on artificial intelligence in 2023 or 2024?” The response included: “It’s important to avoid rushing into legislation solely for the sake of urgency and instead prioritize well-considered, effective, and adaptable regulations that benefit society as a whole.”

Alan S. Inouye is senior director of public policy & government relations at the American Library Association.