Economists at Stanford University have uncovered compelling evidence that artificial intelligence is starting to phase out specific jobs. Nonetheless, the situation is intricate: While AI is taking over positions traditionally held by younger workers in various sectors, seasoned professionals are discovering new prospects.
Erik Brynjolfsson, a professor at Stanford, Ruyu Chen, a research scientist, and Bharat Chandar, a graduate student, examined data from ADP, the largest payroll provider in the US, covering the period from late 2022 to mid-2025, following the launch of ChatGPT.
Their research revealed significant indicators in the data, notably that the rise of generative AI corresponded with a reduction in job prospects for younger workers in industries susceptible to AI automation, such as customer service and software development. These sectors experienced a 16 percent drop in employment for workers aged 22 to 25.
The study presents a complex view of AI’s influence on the workforce. While advancements in AI are often accompanied by forecasts of job cuts, evidence supporting these assertions has been limited. For instance, the relative unemployment rate for young graduates started to decline around 2009, preceding the current AI surge. Some fields, like translation, have, in fact, seen job growth.
Brynjolfsson emphasizes that evaluating the situation based on individual companies or anecdotes is difficult, which is why they opted for a systematic approach.
The Stanford researchers determined that AI’s effects are more closely tied to a worker’s experience and skill level than the nature of the job itself. Skilled workers in sectors adopting generative AI faced less job displacement, with opportunities remaining stable or even increasing slightly. This aligns with what some software developers have indicated regarding the influence of AI—routine tasks, such as coding for API connections, are simpler to automate. The study also implies that AI is eliminating jobs without reducing wages, at least for now.
The researchers took into account various factors, including the Covid pandemic, remote working trends, and recent technology layoffs, discovering that AI’s ramifications endure even when these influences are considered.
Brynjolfsson proposes enhancing AI’s advantages throughout the economy by reforming the tax system to prevent rewarding companies that substitute labor with automation. He also encourages AI firms to create systems that prioritize human-machine collaboration.
In June, Brynjolfsson and Stanford colleague Andrew Haupt contended that AI companies should establish new “centaur” AI benchmarks to assess human-AI collaboration, fostering augmentation rather than automation. Brynjolfsson is convinced that many tasks still gain from human-machine partnerships.
Some specialists predict a future labor market characterized by deeper human-AI collaboration. Matt Beane, an associate professor at UC Santa Barbara who researches AI-driven automation, anticipates that AI will generate a need for work that can be augmented, as managing AI outputs becomes increasingly vital. “We’ll automate as much as we can,” Beane states. “But there will be a growing mountain of augmentable work for humans.”
AI is progressing swiftly, and Brynjolfsson cautions that its effects on younger workers might soon reach more experienced individuals. “We need an early-warning dashboard system to monitor this in real time,” he asserts. “This is a profoundly consequential technology.”
