Almost half of professional employees worry they will be judged for not using artificial intelligence enough at work, according to research suggesting that adoption is becoming an expectation even where its contribution to performance remains uncertain.
A study by the London School of Economics and Political Science (LSE) and consultancy Protiviti found that 48% of respondents feared criticism for insufficient AI use, compared with 20% concerned about being judged for using the technology.
The findings are based on a survey of 3,001 professional employees globally, conducted between March and May 2026 as part of a three-year research programme.
The latest report, Beyond the Technology: Who Really Captures AI’s Gains at Work, found that 74% of respondents used AI in their jobs, compared with 29% in the programme’s initial survey in 2024.
The researchers also examined employees who participated in successive survey waves. Within that group, AI adoption increased from 69% in 2025 to 82% in 2026.
Despite the increase in use, the study found no consistent association between AI adoption and employees’ perceptions of how their performance was viewed by managers.
Training, access to employer-provided AI tools, and the incorporation of AI into established workflows were likewise not consistently associated with higher perceived performance.
Training changes how employees use AI —
Employees who received AI training were substantially more likely to use the technology than those without training.
Adoption reached 93% among trained employees, compared with 50% among untrained respondents. The differences also appeared in the amount of working time employees believed they were saving.
Trained users reported savings equivalent to 22% of working hours, against 13% for those without training. Across the wider sample, employees using AI reported saving an average of approximately 19% of their working time.
These are self-reported estimates rather than independently measured improvements in output, profitability, or the quality of completed work.
The research also identified differences in access to instruction. Around 66% of Gen Z employees had received AI training during the previous year, compared with 40% of baby boomers.
That disparity may influence which employees feel equipped to use the technology and how confidently they evaluate its suitability for particular tasks.
Organisations introducing AI therefore face a question extending beyond the availability of software: whether employees receive training appropriate to their responsibilities and understand when automated assistance adds value.
Greater usage does not necessarily mean that the technology is being applied to the most suitable work. Employees can increase the number of tasks completed with AI without improving the relevance or quality of their contribution.
Who benefits from the time saved —
The findings also raise questions about how organisations use the capacity released by automation.
Employees may be able to complete research, analysis, document preparation, or administrative work more quickly. Whether that time subsequently supports professional development, more demanding work, or additional routine tasks depends on management decisions.
The researchers identify the possibility of workload creep, where efficiency gains are absorbed into higher expectations without corresponding changes in employee recognition or working conditions.
The difference between concern about using too little AI and concern about using it also suggests that workplace expectations may be changing faster than formal performance frameworks.
Employees could feel pressure to demonstrate that they are using AI even where managers have not established which applications produce useful results.
Importantly, the study measures perceived job performance: how employees believe their work is regarded by leaders. It does not establish that AI adoption has no effect on objectively measured performance.
Researchers found stronger associations between perceived performance and human attributes including conscientiousness, resilience, problem-solving confidence, and willingness to learn.
These findings place attention on how businesses evaluate work alongside their investment in AI technology. Adoption rates and reported time savings provide information about usage, but do not establish the value of the resulting work.
The LSE and Protiviti research identifies rapid growth in workplace AI use, substantial differences in training access, and continuing uncertainty over how employees and employers share the benefits of greater efficiency.




