Researchers at Stanford University have presented compelling evidence indicating that artificial intelligence is now beginning to displace certain jobs, with younger workers being replaced in some sectors while older, more experienced workers are finding new opportunities arising.
The study, conducted by Erik Brynjolfsson, a professor at Stanford University, along with Ruyu Chen, a research scientist, and Bharat Chandar, a postgraduate student, analyzed data from ADP, the largest payroll provider in the US, spanning from late 2022 to mid-2025 when ChatGPT was introduced.
Their analysis uncovered significant trends in the data. Notably, the introduction of generative AI corresponded with a decline in job prospects for younger employees in industries previously identified as susceptible to AI-driven automation such as customer service and software development. In these fields, there was a notable 16 percent decrease in employment among workers aged 22 to 25.
This latest research paints a nuanced picture of how AI is impacting the workforce. While concerns about job losses due to advancements in artificial intelligence have been widespread, concrete data supporting these fears has been scarce. For instance, the unemployment rate among recent graduates began decreasing around 2009 well before the current surge in AI development. Surprisingly, sectors like translation that were perceived as vulnerable to AI disruption have actually witnessed job growth in recent years.
Brynjolfsson remarked on the importance of taking a systematic approach to understanding these changes: “It can be challenging to grasp the full scope of what’s happening if we rely solely on individual company anecdotes. Therefore, our aim was to conduct a more comprehensive analysis.”
By meticulously examining payroll records, the Stanford team concluded that the impact of AI hinges more on an employee’s experience and skills rather than the nature of their work. Seasoned workers in industries adopting generative AI technologies were shielded from job displacement, with opportunities either remaining stable or experiencing slight growth. This finding aligns with feedback from software developers who noted that routine tasks like coding for API integration have become easier to automate due to AI advancements. The study also suggests that while AI may be eliminating jobs, it is not causing wage reductions – at least not yet.
Despite considering potential factors like the Covid pandemic, remote work trends, and tech industry layoffs, the researchers confirmed that AI’s influence persists even after factoring in these variables.
Brynjolfsson emphasized that this study offers insights into harnessing the benefits of AI across various sectors. He proposed changes to tax policies to discourage companies from automating jobs and advocated for AI systems that prioritize collaboration between humans and machines.
In a joint paper released in June by Brynjolfsson and fellow Stanford scientist Andrew Haupt, they proposed creating new benchmarks known as “centaur” AI metrics to gauge human-AI collaboration efforts. By emphasizing augmentation over automation, they believe this approach could lead to more effective use of AI technologies. Brynjolfsson highlighted the potential for humans and machines working together to outperform standalone AI systems in many tasks.
Some experts foresee increased collaboration between humans and AI shaping future labor markets. Matt Beane, an associate professor at UC Santa Barbara specializing in AI-driven automation, anticipates a rise in demand for roles focused on overseeing AI outputs as automation becomes more prevalent. He stressed that while automation will continue expanding, there will still be ample opportunities for humans in roles where augmentation is essential.
As AI continues its rapid advancement, Brynjolfsson cautioned that the repercussions felt by younger workers could extend to those with greater experience. He stressed the need for establishing an early-warning system dashboard to monitor these developments closely: “This technology carries significant consequences that require real-time tracking.”