Nearly half of professional employees worry about being seen to use too little AI at work, according to research published on 17 November by The Inclusion Initiative at the London School of Economics and Political Science (LSE) with consulting firm Protiviti. The survey of 3,001 professional employees found that 48% worry about being judged for not using AI enough, against 20% who worry about being judged for using it.

What the survey measured

The report, Beyond the Technology: Who Really Captures AI’s Gains at Work, is the third annual study from the two organisations. It surveyed professional employees globally from March to May 2026. Protiviti is a wholly owned subsidiary of Robert Half.

The headline adoption figure is that 74% of respondents now use AI at work, up from 29% in 2024. In the 2024 survey an AI user was in the minority of the sample. Two years later the non-user is, and the worry about judgement has followed the majority. The share anxious about using AI too little is more than twice the share anxious about using it at all.

Advertisement

HRTech Your brand belongs here. Reach the decision-makers who read HRTech every day. Premium placements across the site and newsletter. Advertise with us

Where the saved time goes

AI users in the survey report saving an average of 19% of their working week, and they reinvest 72% of that saved time in other work. Among users, 82% say AI helps them complete tasks more quickly, 70% report reduced mental effort, 63% say it helps automate manual tasks and 62% say it improves the quality of their work.

Of those who saved time, 65% say they spend at least moderately more time on higher-value activities, and 50% report using saved hours to take on new responsibilities or projects. The release calls the risk “workload creep” and describes it as AI-created capacity being absorbed by greater output and broader responsibilities.

Dr Grace Lordan, founding director of The Inclusion Initiative at LSE, said: “Leaders need to be clear about where AI adds value and how the time it frees should be used. Without that direction, greater efficiency may simply raise the baseline for what employees are expected to deliver.”

The arithmetic for a workforce plan

By our arithmetic, 72% of a 19% saving is close to 14% of the working week going straight back into other work, which leaves about 5% of the week unaccounted for in the release. All of these figures are self-reported by employees. The release does not say how the remaining time is used, and it does not break the results out by country or sector.

Matt Duncan, managing director at Protiviti, said: “Simply absorbing those gains into existing workloads leaves much of AI’s potential untapped.” He added that leaders should think deliberately about where the capacity can create the most value, and listed higher-value work, improving quality and giving employees greater space to solve problems and innovate.

Training tracks adoption, and access is uneven

The survey links AI training to both adoption and time saved:

  • AI adoption is 93% among trained employees and 50% among untrained employees.
  • Where the employer provides AI tools, adoption is 96% among trained employees and 78% among untrained employees.
  • Trained employees report saving 22% of working hours, against 13% for employees without training.

Access to that training differs by generation. In the past year 66% of Gen Z employees received AI training, compared with 40% of baby boomers. The release reports these as associations in survey data. It does not claim that training caused the difference in adoption or hours saved, and employers that train staff may differ from those that do not in other ways the release does not describe.

AI use and how performance is perceived

The finding that cuts against the anxiety is about performance. The researchers found no consistent association between perceived job performance and AI adoption, training, in-house access or integration into standardized workflows. The release defines perceived job performance as how employees believe their work is seen by their leaders.

Higher perceived performance was associated with conscientiousness, resilience, confidence in solving problems and willingness to learn. So 48% of the sample worry about being judged on AI use, while the same study finds that AI use does not track how well employees think their leaders rate them.

Newsletter

Get the week's best tech coverage.

Free. Read by thousands of HR, tech, and business leaders.

In our reading, the two findings together describe employees responding to an expectation they sense and that nobody has stated to them. The release supports the first half of that sentence with its 48% figure. The second half is our inference from Lordan’s call for leaders to be clear, and the survey did not ask employers what they have communicated.

What it means for the HR leader

Three parts of the people function are touched by these numbers.

Performance management

If AI use is part of how people are assessed, employees should be told how. If it is not, they should be told that too. Either statement removes the guesswork behind the 48% figure. We have argued before that HR should write role-level AI expectations before building a skills library, and this survey adds a reason from the employee side.

Learning and development

The gap between 66% of Gen Z and 40% of baby boomers receiving AI training is one an HR team can check against its own records in an afternoon. Enrolment and completion data by age band, job family and location will show whether training reached the people whose work changes most. The time-saved gap between trained and untrained employees, 22% against 13%, gives the L&D budget a number to cite.

Workforce planning

Half of the employees who saved time took on new responsibilities or projects. Where that happens informally, job descriptions and levels drift away from the work people do. Planning teams already struggle to see technology’s effect on roles: a recent Deloitte survey found 16% have a strategy for technology’s workforce impact. Managers are the people who see the drift first, which is part of the case that AI enablement budgets should pay for managers’ time first.

How to test this in your own organisation

The survey covers professional employees and relies on what they report, so treat it as a prompt to measure, and check your own workforce before acting on its percentages. Four checks can run inside existing HR systems:

  1. Add two items to the next pulse survey: whether employees feel judged for using AI too little, and whether they feel judged for using it. Compare the ratio with the survey’s 48% and 20%.
  2. Ask AI users to estimate hours saved in a week and what replaced them. Record whether the replacement work was assigned by a manager or picked up by the employee.
  3. Pull AI training completions by generation and compare the spread with the 66% and 40% in the release.
  4. Review performance criteria and calibration notes for any mention of AI use, and decide whether it stays, with a written definition, or goes.

Teams that run these four checks before the next review cycle will know whether the worry in the LSE and Protiviti sample exists in their own workforce, and which of the three functions should act on it first.

Source: The Inclusion Initiative at LSE and Protiviti, PR Newswire