AI's effect on jobs not what expected, Yale study finds

What the Yale study shows

The Yale researchers evaluated tasks vulnerable to AI and compared projections with actual employment changes. Using a dissimilarity index, they tracked whether job distribution was shifting unusually fast.

To put AI-driven changes in context, the Yale team also examined the workforce shifts during previous technology waves, such as the rise of personal computers in 1984 and the internet surge in 1996. The findings showed that today's AI-driven changes are unfolding at a similar pace, indicating that AI isn't more disruptive than past technological breakthroughs.

The study concluded that jobs labeled as "highly exposed" to AI didn't experience disproportionate declines in employment compared to jobs considered less exposed to AI, such as roles involving tasks that are hard to automate. It also found that overall employment shifts followed patterns consistent with long-term trends, such as the gradual decline of clerical roles and the growth of service-sector jobs.

The study examined recent college graduates, comparing post-graduation trajectories of young adults aged 20-24 with those aged 25-34. Differences were minimal, suggesting AI hasn't fundamentally altered early-career prospects. However, a recent gap in employment outcomes between the two groups -- with adults aged 20-24 measuring six percent lower than those aged 25-34 -- might reflect broader labor market weakness rather than AI's influence.

Overall, the study found no evidence of the widespread job displacement predicted by sensationalized media coverage.

Alarmist headlines miss the point

Fear-driven headlines have painted a stark picture of automation-driven disruption, warning that millions of jobs could disappear. Yet the Yale research highlights an important distinction: Being affected by AI doesn't necessarily mean being replaced by it. Just because certain tasks can be automated doesn't mean companies are deploying AI on a large scale. It also doesn't mean the technology is reliable enough to replace human workers.

"While AI can perform many tasks, there are tasks it can't do, or can't do well," said Isabella Loiaza, a postdoctoral researcher at MIT Sloan School of Management in Cambridge, Mass. "Technology affects work at the task level, so when you aggregate to the occupation level, far fewer jobs are fully automated than many assume."

In her paper, "The EPOCH of AI: Human-Machine Complementarities at Work," Loiaza found that even in extreme simulations where half of all U.S. tasks could be automated, only about 10% of occupations would disappear entirely.

This perspective aligns with real-world observations at IT companies such as BlackLine, a financial operations management platform that has begun to prioritize internal AI initiatives.

"AI is definitely helping employees become more productive," said Sumit Johar, BlackLine CIO. "Whether that leads to reduced headcounts really depends on an organization's priorities -- efficiency, growth or other strategic goals. At BlackLine, we see AI as a tool to help us achieve those objectives, not to cut staff."

However, AI is disrupting roles and reshaping organizational structures, Johar added. For example, many organizations are actively building and expanding generative AI teams, he said.

Graeme Thompson, CIO of Informatica, a data management platform, offered additional context on IT hiring trends. Roles like programmers and coders aren't being deprioritized, he said. "Hiring engineers who write code has been slowing down for a few years, predating the rise in AI tools," he explained. "There's a ton of automation in these roles already, and AI definitely shifts that to another level."

What this means for workers

This gradual evolution, however, doesn't mean workers can relax. While employment numbers remain steady, AI is reshaping jobs beneath the surface, automating routine tasks and leaving humans to focus on oversight, creativity and strategic judgment. Some sectors might already be reducing entry-level roles, potentially limiting opportunities for young professionals to gain experience.

According to the World Economic Forum's 2025 Future of Jobs report, AI could affect nearly 50 million U.S. jobs in the coming years, with about 40% of employers expecting to reduce their workforce in areas where AI can automate tasks. These projections highlight the importance of workforce planning, reskilling and supporting early-career opportunities.

"Even if technology can automate part of a job, it doesn't mean all positions in that occupation go away," MIT's Loiaza said. "Platforms like TurboTax have reduced demand, but there are still plenty of tax preparers. Technology may reduce employment in an occupation, but it rarely eliminates it entirely."

Even if technology can automate part of a job, it doesn't mean all positions in that occupation go away.

Early-career workers are particularly at risk, as AI might automate various entry-level tasks. The Yale study found that as AI automates routine work, recent graduates are increasingly taking on new roles that require broader skill sets.

Informatica's Thompson highlighted the labor implications of this trend: With certain occupations becoming obsolete, it's important to move students away from those careers. That way, he said, graduates won't leave school to find a labor market where their intended job no longer exists.

IT professionals whose jobs involve manually scanning logs, executing quality assurance scripts and writing code should take advantage of the opportunity to reskill, Thompson added.

For workers, the takeaway from the Yale study is one of cautious optimism: Jobs aren't disappearing en masse, but that doesn't mean the tasks that define roles will look the same in the coming years. AI in the workplace is amplifying demand for higher-level skills, such as reasoning, critical thinking and cross-functional collaboration.

Technology changes so fast that what people learn in college can be outdated in just a few years, BlackLine's Johar noted. Hiring managers aren't only looking at a candidate's academic background; Instead, they're focused on core capabilities.

"Adaptability, curiosity, problem-solving, reasoning and communication are what really matter," Johar added. "Those skills will carry you far in a fast-moving, AI-driven economy."

Growing need for AI skills

While the Yale study suggests overall employment has remained steady, other reports show pockets of change. The Wall Street Journal noted IT unemployment rose to 5.7% in early 2025, up from 3.9% a year earlier, largely reflecting AI's effect on routine IT tasks, such as reporting, clerical administration and basic coding.

AI is changing the types of skills that are in demand. Researchers have found that occupations most exposed to AI, such as business analysts and software developers, haven't seen steep declines in employment. Instead, many of these roles are being redefined. Job postings increasingly emphasize AI literacy, data analysis and problem-solving capabilities, signaling that adaptability has become the new baseline competency.

"We see … growing demand for AI-enabled skills but not widespread layoffs," said David Sewell, CTO at Synechron, a global digital transformation firm. He noted a clear shift, with employees who are skilled in AI gradually replacing those who aren't. Roles are evolving to focus more on integrating, managing and innovating with AI throughout the organization.

"Looking ahead, AI literacy will become essential for all IT staff," Informatica's Thompson added. "While filling skill gaps may require funding or resource reallocation, AI is expected to create more jobs overall, though they may be different roles."

Looking ahead, AI literacy will become essential for all IT staff.

Thompson also noted that the increasing need for workers skilled in AI extends to cybersecurity. "We need people who can protect against new risks from generative AI, including sophisticated automation, personalized attacks and exposure of confidential data," he said.

Joe Locandro, CIO of software company Rimini Street, added that advancements in AI-powered low-code and no-code applications mean less skilled workers can develop scripts. IT departments, therefore, have an increasing demand for skills related to managing the risk of unsafe AI-generated code.

Locandro also emphasized that generative AI is accelerating the value of data as a core business asset. This translates to an increasing demand for IT pros who understand and can curate data to drive enterprise reporting and workflows.

This broader shift in responsibilities is also influencing hiring trends across the enterprise. As organizations scale AI initiatives, new technical and governance-focused roles are emerging.

As AI adoption grows, certain roles are seeing higher demand, Sewell said. These include AI and ML engineers, MLOps specialists who deploy and monitor models, and solution design engineers who integrate AI into modern systems.

He added that data engineers and platform specialists are also crucial in maintaining clean, high-quality data and scalable pipelines. Additionally, finance-focused data scientists are increasingly needed to tackle challenges such as risk, pricing and fraud, all while adhering to strong data governance frameworks.

Taken together, these trends point to a workforce in transition. The effect isn't widespread job loss, but a transformation of work itself, as roles increasingly blend human judgment with machine assistance.

A story of continuity -- for now

AI's lack of dramatic disruption in the job market today doesn't mean disruption isn't coming. Past technological changes often took decades to reshape labor markets. AI might follow a similar trajectory of gradual adoption at first, followed by sharper shifts as tools improve and companies reorganize work around them.

The current anxiety toward AI mirrors historical cycles, Johar said. "Since the Industrial Revolution, people have worried that machines would take their jobs. What's different now is the pace. Change is happening so fast that people feel unprepared to deal with it."

Loiaza emphasized the importance of human agency and collaboration in shaping the future of AI in the workplace. "Even as AI gets smarter, there's a huge difference between a technology being capable of a task and humans deciding we want to use it," she said.

"AI should augment workers and support collaboration, not isolate people," she added. "Protecting human decision-making and fostering teamwork has to remain central."

The Yale study calls for nuance. Rather than fearing mass job losses, it suggests that policymakers and employers should track real AI adoption, invest in reskilling programs and support early-career opportunities so younger workers aren't left behind.

"Labor markets don't shift overnight," Loiaza said. "The real question isn't whether AI takes jobs, but how it changes them and how ready we are to shape that change."

For now, the story of AI and jobs remains one of continuity, not collapse -- a reminder that the future of jobs isn't being erased, but rewritten, one evolution at a time.