60% of the employees surveyed by IBM say they’re worried that artificial intelligence (AI) is eroding their skills, with critical thinking cited most often among the capabilities in decline. The study, conducted by the IBM Institute for Business Value together with Oxford Economics, gathers responses from 1,500 executives responsible for workforce strategy and 8,800 employees across different countries and industries.
IBM’s study on AI and employment in 30 seconds
- 60% of employees fear AI is eroding their skills, especially critical thinking.
- 71% of HR executives prioritize overseeing, validating, or overriding AI outputs, versus 38% of employees.
- 46% of organizations don’t include the head of HR in defining AI strategy.
- 80% of HR leaders see additional work needed to validate outputs, fix errors, or manage exceptions.
- Organizations that define which tasks are human, AI-assisted, or AI-executed report 18% less risk and 20% higher quality.
The report, titled Designing the Thinking Organization, examines how work is changing as companies adopt AI systems and agents capable of taking on cognitive tasks. Its scope goes beyond measuring productivity: it also looks at which capabilities employees need to keep, how jobs and processes are being redesigned, and what role HR leaders play.
The research was conducted between April and June 2026. The executive survey included 1,500 chief human resources officers (CHROs) and other executives responsible for workforce strategy across 21 geographies and 23 industries. The second survey gathered responses from 8,800 full-time employees in 28 countries. The report also draws on interviews with HR leaders conducted between April and August 2026.
The gap between what companies expect and what employees prioritize
One of the findings IBM highlights most is the gap between the capabilities HR leaders consider necessary and the ones workers actually prioritize.
71% of CHROs consider it essential for employees to know how to oversee, validate, or override AI-generated outputs. Among employees, that figure is only 38%. The gap also shows up in judgment under ambiguous situations: 48% of CHROs include it among the relevant capabilities, compared with 29% of employees.
There’s greater alignment on critical thinking and problem definition. 57% of HR leaders and 49% of employees consider these capabilities important in the AI era.
The report itself identifies a tension here: employees recognize they’ll need to adapt to AI, but may be placing less weight than organizations expect on the capabilities needed to evaluate its outputs. IBM notes this could affect both experienced professionals, who might exercise certain capabilities less, and early-career workers who haven’t yet had time to develop them.
The concern isn’t just a perception held by executives. Among employees worried about skill erosion, three out of four say AI has already begun to erode at least one of their capabilities. Critical thinking is the skill they most frequently identify as having deteriorated.
The report also flags another issue: only 23% of organizations are increasing entry-level hiring, while 77% are keeping it flat or reducing it. IBM suggests that if certain early-career tasks disappear due to automation without creating other learning paths, companies may need to find new ways for professionals to build experience — a concern that echoes what cloudnews.tech has reported about AI reshaping entry-level hiring at consulting firms.
Among the alternatives mentioned are job rotations, early client exposure, supervised decision-making, mentoring from experienced professionals, and decision simulations.
Work is changing faster than organizations are
The study estimates that 29% of workflows already incorporate AI agents, and puts that share at 40% by 2028. Yet only 26% of organizations largely define which parts of the work are human-led, AI-assisted, or AI-executed.
The gap also shows up in workforce reorganization. Only a quarter of companies dynamically and continuously manage the reshaping of their workforce as work itself changes.
According to the report’s data, organizations that clearly define their workflows as human-led, AI-assisted, or AI-executed report an 18% reduction in risk and a 20% improvement in quality. IBM notes in its methodology that the relationships between these capabilities and outcomes come from cross-sectional data and should be read as observed associations, not proof of direct causation.
The report also finds differences in employee trust. When human judgment is built into work design, 62% of CHROs say employee trust in AI-supported decisions is increasing. When that component isn’t in place, 57% see trust declining.
Accountability is another issue the report identifies. 43% of employees say that when something goes wrong with AI, responsibility ends up falling on them. Among executives, 36% believe unclear accountability complicates the rollout of these technologies.
The report proposes shifting part of that accountability from the org chart to the workflows themselves. That means defining who validates a decision, who handles an exception, who can halt a process, and who has the authority to override an AI-generated recommendation.
Productivity gains also create invisible work
Adopting AI doesn’t necessarily eliminate every task tied to a process. 80% of the CHROs surveyed believe AI adoption creates “invisible” work for employees, such as validating recommendations, fixing errors, providing context, or managing exceptions.
42% of employees say AI is increasing their workload or that part of the work they do is going unrecognized.
The study therefore argues that performance evaluation systems need to account for that work too. Measuring output volume alone can leave out the activities needed to make an AI tool produce reliable results.
Another notable finding is that only 42% of organizations say they mainly reinvest AI-driven productivity gains into innovation or reskilling. The rest split between savings and reinvestment, put the gains toward cost or profit reduction, or have no clear path for turning freed-up time into new opportunities.
IBM also links advanced AI adoption to strong change-management capability. Organizations that combine both factors report up to 73% higher revenue growth and an 11-percentage-point improvement in operating margin compared with peers, according to the analysis cited in the report. This is an association drawn from IBM research, not a causal relationship demonstrated by this study.
HR is still just getting started with its own AI
The study finds a paradox in the role assigned to HR. CHROs are expected to prepare organizations for change, yet the HR department itself still uses AI in a limited way.
72% of organizations report limited or no use of AI within HR. Breaking that down, 16% use no AI in HR, 26% use built-in features of their existing platforms, 30% run pilot projects for select use cases, 22% have scaled AI to multiple processes, and only 6% describe themselves as an AI-native HR function at scale.
Internal capabilities also vary. Only 13% of executives rate their HR teams’ AI literacy as high, while 16% say the same about measuring AI performance and 20% about managing the change associated with its adoption. AI governance and ethics rank highest, at 31%.
The report suggests HR use its own processes as a testing ground to define rules for transparency, human review, and accountability before extending AI to more sensitive decisions about employees.
This matters especially because HR handles decisions related to hiring, promotion, compensation, and performance. IBM notes that 94% of organizations it classifies as Thinking Organizations guarantee transparency and explainability when AI affects an employee’s career, compared with 62% of the rest, a theme also present in IBM’s separate research on how CEOs are redesigning leadership for the AI era.
The study defines a Thinking Organization as one that combines a set of capabilities across five areas: maintaining human accountability alongside AI, redesigning work and the workforce, developing skills while preventing their erosion, nurturing employee experience and culture, and using HR as a testing ground for trustworthy AI.
In its methodology, IBM builds an index from ten capabilities and classifies 375 organizations in the top quartile as Thinking Organizations. The report notes this classification is based on company responses and statistical associations, not an external certification.
The study’s overall takeaway is that adopting AI isn’t just about introducing new tools. It also requires deciding which tasks still need human judgment, how professional experience is developed, and what happens to the time companies free up through automation. In that reorganization, IBM places HR in a position that, according to the study itself, many organizations still haven’t fully incorporated into their AI decisions.
Frequently Asked Questions
How many employees took part in IBM’s study?
The study gathered responses from 8,800 full-time employees across 28 countries. It also included 1,500 CHROs and other executives responsible for workforce strategy from 21 geographies and 23 industries.
Which skill do employees think is eroding the most?
Critical thinking is the capability most frequently mentioned by employees worried about skill erosion. 60% of all employees surveyed expressed concern about losing capabilities because of AI.
What’s the gap between CHROs and employees on human judgment?
71% of CHROs consider it important to know how to oversee, validate, or override AI outputs, compared with 38% of employees. On judgment under uncertainty and ambiguity, the figures are 48% and 29%, respectively.
What role does IBM propose for HR?
The report suggests HR take part in designing AI-driven workflows, define accountability and rules for people-related decisions, and use its own function to test AI systems with transparency, oversight, and accountability mechanisms.
