CNN  —  About 10% of US workers are in jobs that face the greatest risk of disruption from rapidly evolving artificial intelligence, according to a White House analysis shared first with CNN. The report represents the most in-depth analysis to date by the White House on the impact of AI on the US workforce. It finds that workers with less education and lower income are especially exposed to AI, raising the risk that the technology could amplify inequality. The findings are part of the Council of Economic Advisers’ annual Economic Report of the President, which devotes an entire chapter to AI and how policymakers should respond. “If we were thinking about this in health terms we would be saying: Who is most likely to catch the virus and what can we do to vaccinate them?” Jared Bernstein, chair of the White House Council of Economic Advisers, told CNN in an interview. The report shows how White House officials are thinking deeply about AI and what can be done now to guide its development in a way that ultimately benefits workers. “There have been developments in social media and technology that governments have underattended to as they evolved. We won’t let that happen with AI,” Bernstein said. Bernstein said the White House is in discussions with labor unions — especially in the service industry but also in manufacturing — about being ready for AI. As with any kind of new technology, experts believe AI will complement the work of some people by making them more productive. Other jobs could get wiped out. The White House report contains numerous caveats, including cautioning that the ultimate impact of AI on workers may change as the technology and its capabilities evolve. Generative AI is already able to do some tasks that in the past only humans could, such as writing humorous stories, generating realistic images and crafting song lyrics. 20% of workers are in high-exposure AI jobs White House economists analyzed the impact on workers by identifying 16 work activities with high exposure to AI. Researchers then tried to determine which occupations have AI-exposed activities that are central to the job itself. The report found that about 20% of workers are in these high-AI exposure occupations. That finding is similar to one from Pew Research Center that concluded 19% of American workers in 2022 were in jobs most exposed to AI. “Workers in such occupations are plausibly the ones most likely to be affected by AI, whether positively through complementarity, or negatively through substitution or displacement,” the CEA report said. But to drill down on which roles are most vulnerable to being displaced, researchers broke out jobs by how difficult the tasks are. “Labor-substitution is easiest and cheapest in situations where complexity and difficulty are low,” the report said. In other words, the more complex the job, the safer it is. And vice versa. Displacing versus complementing workers The researchers found that 10% of workers have high AI exposure and low performance requirements. The report said those workers “perform the tasks that are most likely to change as a result of AI.” It did not identify specific occupations or industries that fall into this category. However, researchers stressed that this does not mean 10% of workers “will inevitably lose their jobs” as the implications for workers may be “quite nuanced.” “Most jobs remain a collection of tasks of which only a portion can be automated,” the report said. “AI may allow humans to focus on other tasks, fundamentally changing their jobs without reducing the use of their labor.” As an example, the White House economists point to how even if AI eventually allows school buses to drive themselves, the school bus driver’s job won’t necessarily be killed. “Children may still need someone on the bus to watch them, ensure they behave and ensure they enter and exit safely,” the report said. “AI-led automation might fundamentally change the school bus driver’s job, but it is unlikely to eliminate the job.” The report notes that airlines still have pilots even though autopilot systems have been around for more than a century. “The wrong place to start a discussion of AI is to assume without question it will massively displace workers. That’s not the history of technology in the workplace,” Bernstein said. AI could worsen inequality Still, some workers are more exposed to AI than others. The White House report found that 14% of high school graduates lacking four-year degrees have jobs that have high-AI exposure but low performance requirements. By comparison, only 6% of workers with a bachelor’s degree fall into that higher risk bucket. Likewise, women are more likely to have high AI exposure with low performance requirements, the report said, “suggesting that women may be at higher risk of displacement.” Lower-income workers may also face greater risks from AI disruption. The report found that nearly 40% of workers in the third decile of earnings have high-AI exposure and low performance requirements. High-earnings also have high exposure to AI but their job requirements are more complex. The findings “suggest that AI may be a skill-biased technology, increasing relative demand for workers with high levels of education in high-earning occupations,” the report said. “They also suggest that AI could exacerbate aggregate income inequality if it substitutes for employment in lower-wage jobs and complements higher-wage jobs.” The International Monetary Fund warned earlier this year that in most scenarios, AI will “likely worsen overall inequality, a troubling trend that policymakers must proactively address to prevent the technology from further stoking social tensions.” However, the White House report stressed that it’s too early to definitively say AI will worsen inequality — in part because this risk may influence policy. Bernstein said the Biden administration wants to implement policies that will lower the risk that AI displaces workers. “There is always technology afoot, but society has the level of inequality that policymakers are willing to accept,” he said. “We’re not going to let technological developments determine inequality.”