In March, the U.S. labor market recorded 178,000 new jobs, little changed from the previous month, according to the Bureau of Labor Statistics.
The anemic growth in job openings comes amid volatile policy changes by the White House, rising energy prices due to the US and Israel’s war with Iran and, according to recent research, AI disruptions in the labor market.
Proponents of AI and large language models have claimed that the technology will spark an economic boom, thanks to the promise of advances in efficiency.
But as AI becomes more integrated into daily business operations, there is a growing chasm between that promise of growth and efficiency and what is actually happening.
AI slows down job growth
On March 6, venture capitalist and Netscape co-founder Marc Andreessen said on X that fears about AI displacing jobs were overblown.
He also published a Business Insider article stating that, at least in technology, job openings are increasing. Citing data from TrueUp, a tech job tracker, Business Insider said job openings at tech companies have doubled to 67,000 since 2023.
But vacancies don’t necessarily translate into hiring. According to the Bureau of Labor Statistics, most of the job growth in March did not occur in the technology industry. Of the 178,000 new jobs added in March, healthcare employed 76,000, construction grew by 26,000, transportation and warehousing added 21,000 and social assistance employment increased by 14,000.
While the report does not have a single section tracking the tech industry, related services, such as IT infrastructure providers and web search portals, saw a decline of 1,500 jobs, or almost no change, respectively. Computer systems design and related services lost 13,000 jobs.
Related: Jack Dorsey’s Block to Cut 4,000 Jobs in AI-Powered Restructuring
In fact, AI has eliminated 16,000 jobs per month over the past year, according to a recent Goldman Sachs report, cited by Fortune. In particular, AI has caused a collapse in hiring for entry-level positions. A 2025 study by SignalFire found that new graduate hiring had decreased by 50% compared to pre-COVID-19 pandemic levels.
“The door to technology was once wide open for new graduates. Today, it is barely open. The industry’s obsession with hiring bright-eyed graduates right out of college is colliding with new realities: smaller funding rounds, shrinking teams, fewer new graduate programs, and the rise of AI,” the SignalFire study states.
This disruption could have repercussions in the distant future. According to Goldman Sachs, “AI-driven displacement could impose lasting costs on affected workers, worsening labor market outcomes for several years.”
“A key mechanism behind these worse outcomes is occupational downgrading. Workers displaced by technology are more likely to move into more routine occupations that require fewer analytical and interpersonal skills, likely because the same technological changes that eliminated their jobs also eroded the value of their existing skills,” they continued.
These job losses are justified by the theory that AI will, at a minimum, make workplaces more productive. But even that is not a fact.
The reality of the use of AI clashes with the expectations of senior management
Executives continue to overwhelmingly support AI. According to Harvard Business Review, 80% of leaders report weekly use of AI, and 74% report positive returns on first implementations.
But the workers do not feel the same. A study by human resources consultancy Mercer found that, for 43% of employees, their work is more frustrating.
A major problem is the number of errors that generative AI produces. “For every 10 hours of efficiency gained through AI, nearly four hours are lost fixing its performance,” a Workday report states.
AI can also be used to offload work to coworkers in what Harvard Business Review researchers have called “sloppy work,” that is, “content that appears polished but lacks real substance, offloading cognitive work to coworkers.”
They said that “41% of workers have encountered these types of AI-generated results, costing almost two hours of rework per instance and creating downstream productivity, trust and collaboration issues.”
According to Workday, only 14% of respondents said they “consistently achieve net positive results from using AI.”
Part of the gap between executives’ understanding of AI and the reality at the production level can be explained by the technology itself.
According to Harvard Business Review, “Top leaders tend to use AI for high-level synthesis, strategic writing, and decision support—tasks where the technology works well, so current capabilities tend to benefit their work.”
For more complicated daily operations, such as “workflows built over years, teams with uneven technical comfort, results that must always be correct, not just fast,” it doesn’t work as well.
“When the tool works, both groups understand and reap the benefits. When it fails, usually only one of them has to deal with the consequences.”
Brian Solis, head of global innovation at enterprise AI company ServiceNow, said this division has created an “AI tax,” meaning “More control. More rework. More anxiety. Faster pace. Deterioration of AI. Less trust.”
Andreessen may not believe the AI job cuts narratives are real, but OpenAI does. The AI company has recognized the impact that technology has on employment, and has even published a series of policy proposals to address it.
The list contains ideas that are “intentionally early and exploratory” that serve as “a starting point for discussion upon which we invite others to build.” It includes proposals to expand health coverage, retirement savings and the establishment of a new industrial policy agenda.
Far from Andreessen’s optimism, OpenAI’s proposal included a warning: “Unless policies keep pace with technological change, the institutions and safety networks needed to navigate this transition could be left behind.”
Magazine: Asia Express: Bitcoin phantom checks, China tracks blockchain taxes
