![\"promo](\"https:\/\/www.newsbeep.com\/us\/wp-content\/uploads\/2026\/10\/UXDD4BKPO5BJVFTSEM3U6OG5RE.png&high_res=true&w=2048.webp\"\/)
(Illustration by Yan Wu\/The Washington Post; iStock) ![\"\"](\"https:\/\/www.newsbeep.com\/us\/wp-content\/uploads\/2026\/10\/5520d705-fd7e-4f86-a18e-97003094482d.png&h=160.png\")
<\/p>\n(Illustration by Yan Wu\/The Washington Post; iStock) <\/p>\n
October 13, 2026 at 7:00 a.m. EDTToday at 7:00 a.m. EDT<\/p>\n
Ask a software engineer with 10 years of experience whether artificial intelligence is taking their job, and they\u2019ll probably shrug. They still have plenty to do, overseeing custom-made bots debugging code even as the engineers sleep. Ask someone who graduated this spring, however, and you\u2019ll hear a different story: a mountain of applications, a handful of interviews and zero offers.<\/p>\n
This is a common pattern in AI adoption. People with established skill sets are doing just fine, while younger workers face greater headwinds<\/a>. Yet this entry-level pain has not been uniform across all industries. Understanding where it is being felt \u2014 and where AI is expanding jobs \u2014 will help young people choose which career ladders are worth climbing.<\/p>\n Stanford University\u2019s Digital Economy Lab<\/a> has been tracking these nuances using payroll records from the HR firm ADP, which covers millions of American workers. It used that data to build an index showing employment growth across occupations, which it broke into five groups based on exposure to AI.<\/p>\n Though all of those groups have grown since OpenAI released ChatGPT in 2022, the picture looks far worse for workers between the ages of 22 and 25. For that age group, the most AI-exposed occupations have shrunk. (The Washington Post has a content partnership with OpenAI.)<\/p>\n In fact, experienced workers in those same occupations showed no similar gap. That indicates the shrinking of the most-impacted industries comes almost entirely from declining head counts, not from people being fired.<\/p>\n This can do long-term damage for young people in impacted fields. Entry-level work isn\u2019t only how young people earn; it\u2019s how they learn. Grunt jobs often double as apprenticeships. In software development, that includes managing ticket queues and debugging code, tasks where AI unfortunately excels. So companies are cutting entry-level head counts while keeping senior employees who already served their time. <\/p>\n Not every field has lost that bottom rung, however. The jobs holding up are the ones built on knowledge that never got written down. For instance, demand for entry-level home health aides is rising.<\/p>\n That field illustrates some of the characteristics of a job that cannot be automated. Home health aides, of course, must be physically present to offer care. They must also maintain a high level of accountability and trust that simply cannot be compressed into a prompt.<\/p>\n Similar themes can be found throughout OpenAI\u2019s jobs transition framework<\/a>, which projects which careers will grow and which will shrink based on four questions: Can AI do a big chunk of the tasks? Is it already doing so? Does a human still need to deliver the work and take the blame if things go wrong? And if the service gets cheaper, will demand surge enough to create new jobs?<\/p>\n Keep in mind that, like other frontier labs, OpenAI\u2019s business model depends on convincing companies that they can save some of what they spend on employee salaries by adopting its technology, so the firm might have a bias that causes it to overstate AI\u2019s impact. Nevertheless, its data is helpful to at least conceptualize how much the labor force will change. <\/p>\n It finds that 18 percent of jobs land in the high-automation-risk bucket. But almost half are less likely to change in the near term, particularly if they involve physical tasks, and 12 percent of jobs could grow as AI improves the quality of products and services or makes them more affordable or accessible, thereby increasing demand.<\/p>\n Type in an occupation and see how AI could affect it: <\/p>\n Labor demand elasticity (how much demand rises as costs fall) \u2192<\/p>\n An elasticity of 1.0 means that for every 1% decrease in service costs, demand for that service will increase by 1%.<\/p>\n It\u2019s not yet clear how the market will absorb whatever productivity boost comes from AI. Keeping roughly the same number of employees could let a company do a lot more. For example, a YouGov poll<\/a> of more than 1,000 British teachers, commissioned by the British Educational Training and Technology Show, found that roughly 80 percent now use AI at work. About half said it lightened their workload, yet only about a third said they were working fewer hours. They reported redirecting that saved time toward addressing backlogs and fulfilling additional responsibilities from their employers.<\/p>\n