The US labor market added 178,000 jobs in March, according to the Bureau of Labor Statistics — a modest gain that raises questions about whether corporate claims of AI-driven hiring and productivity are materializing across the economy.
Most of the new jobs were outside core tech. Healthcare led with about 76,000 additions, construction added 26,000, and transportation and warehousing gained 21,000. By contrast, tech-linked categories showed little improvement: computing infrastructure and web-search portals were largely flat, and computer systems design and related services shed roughly 13,000 positions.
That mix contrasts with public assertions of a tech hiring recovery. Some industry leaders have pointed to rising openings at tech firms, but openings have not consistently turned into hires. March’s largest gains occurred in non-tech sectors, while many digital services remained steady or declined.
Entry-level roles appear particularly strained. A Goldman Sachs analysis cited in recent reporting estimated AI displaced roughly 16,000 jobs per month over the past year. A 2025 SignalFire report found new-graduate hiring in tech about 50% below pre-pandemic levels, blaming smaller funding rounds, leaner teams, fewer university hiring programs, and greater AI use. SignalFire summed up the shift this way: the once-wide door for new grads is now barely ajar.
Goldman Sachs also warned that displaced workers often move into more routine roles, which can erode their skills and weaken long-term career prospects. That dynamic has intensified debate over who bears the costs of technological change, even as some executives emphasize potential long-term benefits.
Executives and managers often report positive early results from AI. Harvard Business Review found roughly 80% of senior leaders use AI weekly and 74% report favorable returns from initial deployments. But many frontline workers describe a different experience. Mercer reports 43% of employees say their jobs have become more frustrating since AI was adopted. Workday’s data suggests for every 10 hours of claimed efficiency gains, nearly four hours are spent correcting AI output. HBR highlighted a phenomenon it called “workslop” — polished-looking but shallow AI-generated content — which 41% of workers reported seeing; each instance adds about two hours of rework. Workday also found only about 14% of respondents consistently achieve net-positive outcomes from AI use, pointing to frequent errors, added review time, and low trust.
Part of the gap may be how tools are applied. Senior leaders tend to use AI for strategy, drafting, and synthesis — areas where models often perform acceptably — while routine operational tasks that require steady accuracy see less reliable results. Brian Solis of ServiceNow described the extra burden as an “AI tax”: more checking, more rework, and more anxiety for workers.
Recognizing these disruptions, OpenAI has proposed policy measures — expanded healthcare coverage, stronger retirement savings support, and an industrial agenda — to help manage transitions. The company warned that without policy measures keeping pace with technological change, institutions and safety nets needed to navigate the shift could lag behind.
In short, March’s modest job gain underscores a complex picture: companies report widespread AI adoption and some early returns, but hiring patterns and worker experiences suggest the benefits are uneven and that broader economic and policy questions remain unresolved.