One in six STEM professionals surveyed across five countries reported seeing colleagues leave their organisations during the previous year after artificial intelligence changed those colleagues’ work beyond recognition, according to research published by SThree.
The findings, released on 20 October 2026, come from an online YouGov survey of 4,126 employees working in science, technology, engineering and mathematics-related occupations in the United Kingdom, Netherlands, Germany, United States and Japan.
Fieldwork took place between 10 July and 1 September 2026, with respondents asked about changes to technical roles, career opportunities, workloads and the departure of specialist employees.
Global STEM workforce consultancy SThree reported that 16% of respondents had observed colleagues leaving after AI changed their work beyond recognition.
This is a measure of respondents reporting colleagues’ departures. It does not mean that 16% of STEM employees resigned, nor does it establish an overall workforce attrition rate attributable to AI.
The finding also depends on respondents’ understanding of why colleagues left their organisations. It is not based on independently verified employment records or interviews with every person whose departure was reported.
SThree identified two additional reasons that respondents associated with colleagues leaving. Fourteen per cent said they had observed departures for AI-related reasons involving reduced career progression, while 13% reported colleagues leaving because of workloads regarded as unsustainable.
These measures may overlap. A respondent could have observed more than one type of departure, or identified several factors affecting the same colleague. The percentages cannot therefore be added together to calculate a combined AI-related resignation rate.
The findings form part of SThree’s research into how technical employees perceive changes taking place within their organisations as AI becomes more widely used.
The survey also examined recruitment difficulties. Forty-five per cent of respondents said their organisation had struggled to replace specialists who left during the preceding year.
That result refers to replacing departing specialists generally. It is not a finding that 45% of organisations were unable to recruit replacements specifically for employees who left because of AI.
Specialist recruitment was already a focus of SThree’s business before the survey. The company operates in markets requiring expertise in engineering, life sciences, technology and other technical disciplines, including positions where candidates may need experience of particular systems or processes.
Chief executive Timo Lehne said employers were competing for specialist expertise while technological change was altering the work performed by some of those employees.
He argued that the retention of experienced workers could become more difficult where the responsibilities, development opportunities or working conditions associated with their jobs changed substantially.
That assessment is SThree’s interpretation of the findings. The survey does not establish how many individuals left particular industries or whether the employees concerned subsequently joined competitors, moved into other occupations or withdrew from employment.
Respondents were also asked about the pace of technological change. Thirty-seven per cent believed their organisation was introducing technologies such as AI too quickly to consider fully the effects on employees’ roles.
The measure describes respondents’ perceptions of implementation. It does not establish that the organisations concerned lacked formal change management procedures or that particular deployment decisions breached employment obligations.
A further 32% reported that senior employees were undertaking work previously performed by junior colleagues as AI altered entry-level responsibilities.
This suggests that some respondents are observing changes in how tasks are distributed between employees at different career stages. The research does not specify the number of junior positions removed, the duration of those changes or whether employers introduced alternative training arrangements.
The distinction between an employee’s role and the tasks within that role is relevant when examining the findings. A change in the work undertaken by staff does not necessarily mean that the position itself has been eliminated.
SThree’s survey records employees’ observations of changes in responsibilities and departures, rather than providing a direct comparison of headcount before and after AI implementation.
The five countries covered are significant markets for scientific research, engineering and technology employment. However, the combined results should not automatically be applied to individual countries or technical occupations without examining their respective samples.
The company said the markets included in its research account for a substantial share of international research expenditure and patent activity. That provides context for its selection of countries but does not establish that the survey represents the worldwide STEM workforce.
SThree has not published a verified count of AI-related resignations, a financial estimate of the resulting recruitment costs or a causal analysis isolating AI from other factors influencing employee decisions.
The findings instead document several reported workplace experiences: colleagues leaving after substantial changes to their roles, concerns about progression and workloads, difficulties replacing departing specialists and perceptions that technology is being introduced too quickly.
Lehne said movement between organisations could also make experienced professionals available to new employers. That is a potential recruitment opportunity identified by SThree, rather than an outcome quantified by the survey.
The research adds detail to employees’ perceptions of how AI is affecting technical work across five major markets. Its central 16% finding remains a measure of respondents who reported observing colleagues leave for the specified reason, not the proportion of the STEM workforce that has resigned because of AI.
You must be logged in to post a comment.