90% of firms see no measurable AI impact on productivity or jobs: Study

Despite years of investment and growing adoption, artificial intelligence has yet to deliver measurable improvements in productivity or employment outcomes at most organisations, according to a working paper published by the National Bureau of Economic Research (NBER).

The study, revised in March 2026, surveyed nearly 6,000 CEOs, CFOs and senior executives across the United States, United Kingdom, Germany and Australia. Its central finding is stark: more than 90% of firms reported no measurable impact on employment, while 89% reported no measurable effect on labour productivity over the previous three years.

The findings challenge a dominant narrative in boardrooms and financial markets that AI is already transforming workforce efficiency and organisational performance.

Adoption is growing, but results remain limited

The research, authored by Ivan Yotzov, Jose Maria Barrero, Nicholas Bloom, Philip Bunn, Steven J. Davis and other contributors, found AI usage is already common across surveyed organisations.

According to the study:

• 69% of firms are actively using AI
• More than two-thirds of executives report using AI regularly
• Average executive usage stands at approximately 1.5 hours per week
• Around one-quarter of executives say they do not use AI at work at all

While adoption rates are rising, the researchers found little evidence that these investments have translated into measurable improvements in workforce productivity or staffing levels.

Labour productivity in the study was measured using sales generated per employee.

Executives expect future gains

Although current results appear limited, executives remain optimistic about AI's long-term impact.

Survey respondents forecast:

• 1.4% growth in productivity over the next three years
• 0.8% increase in output
• 0.7% decline in employment

The contrast between present outcomes and future expectations mirrors a historical debate around technological change.

As reported by Fortune, the findings echo economist Robert Solow's famous observation from 1987 that the computer age was visible everywhere except in productivity statistics. Fortune also cited Torsten Slok, chief economist at Apollo, who noted that AI remains highly visible across industries but has yet to appear meaningfully in employment, productivity or inflation data.

Workforce decisions are moving ahead of measurable returns

The findings arrive as organisations increasingly reference AI when announcing restructuring programmes and workforce reductions.

According to reporting by TechCrunch, companies including Intuit, Atlassian and Monday.com have linked hiring decisions or workforce changes to AI initiatives.

Separately, the Financial Times reported that organisations including Oracle, Salesforce, Lufthansa, Accenture and Standard Chartered have referenced AI as part of job-cut decisions.

The NBER findings suggest a disconnect between these organisational moves and measurable business outcomes.

While companies may anticipate future efficiencies, the survey indicates those benefits have yet to materialise at scale across most organisations.

Mixed evidence from early adopters

Several high-profile companies have reported operational improvements linked to AI, though the relationship between automation and workforce reduction remains complex.

Salesforce said in April that its Agentforce platform had handled 2.6 million customer conversations, achieving a 63% resolution rate. The company also reported that hundreds of support engineers had been redeployed rather than replaced.

Fortune later reported comments from CEO Marc Benioff, who said Salesforce had reduced its customer support workforce from 9,000 employees to 5,000.

Meanwhile, Amazon continued to reshape parts of its workforce even as it expanded AI investments. According to reporting by Computerworld and other outlets, Amazon confirmed approximately 16,000 job cuts in January 2026. Reuters later reported reductions within parts of the company's artificial general intelligence operations.

These developments highlight a broader trend in which AI investment and workforce restructuring are occurring simultaneously, though direct causal links often remain difficult to establish.

Customer experience remains a critical consideration

The debate over AI's workforce impact extends beyond productivity metrics.

Klarna became one of the most closely watched examples after disclosing that its OpenAI-powered assistant handled 69% of customer service chats during the 12 months ended 30 June 2025.

According to the company's SEC filings, the system performed work equivalent to more than 700 full-time agents and generated approximately $39 million in cost savings during 2024.

However, Bloomberg reported in May 2025 that Klarna CEO Sebastian Siemiatkowski later acknowledged the company's focus on customer-service cost reductions had gone too far.

  TechCrunch subsequently reported his comments at SXSW London, where he emphasised that customers should continue to have access to human support when needed.

The episode illustrates a growing challenge for employers seeking to balance automation, efficiency and customer expectations.

Research points to a widening perception gap

The NBER findings align with other recent studies that question the immediate business impact of AI.

According to PwC's 2026 Global CEO Survey, which surveyed 4,454 chief executives across 95 countries:

• 56% reported neither revenue nor cost benefits from AI
• Only 12% reported achieving both revenue and cost benefits

A separate survey by Section, covering 5,000 white-collar workers, revealed a significant difference between executive and employee perceptions.

The survey found:

• 40% of non-managers reported no weekly time savings from AI
• 19% of executives said AI saved them more than 12 hours per week
• Only 2% of non-managers reported comparable time savings

The findings suggest AI's perceived value may vary considerably depending on organisational role and proximity to strategic decision-making.

The productivity question remains open

The NBER paper does not conclude that AI lacks value. Instead, it suggests that measurable business returns remain limited despite widespread experimentation and investment.

For organisations accelerating AI adoption, the research raises a critical question: whether current expectations are running ahead of demonstrable results.

As businesses continue to invest heavily in AI infrastructure, tools and workforce transformation, the challenge for leaders may increasingly shift from deploying AI to proving its impact.