A significant share of accounting professionals remain cautious about the reliability of generative AI (genAI), even as they say it is boosting productivity, according to new research from the Institute of Chartered Accountants of Scotland (ICAS).

The findings are set out in a new report, Generative AI and Professional Judgement in Accounting, released under the ICAS Shaping the Profession initiative.

According to the survey, around two-thirds of participants (72%) feared that genAI might generate incorrect output or support flawed decisions, pointing to the need for active human review and control.

This is despite 74% of respondents saying genAI helped them complete certain tasks more quickly.

However, only a quarter (22%) of them reported that these time savings had turned into a marked rise in their overall output.

The year-long study was carried out by researchers from the Alliance Manchester Business School and Aston Business School.

It examined how genAI is being introduced and used at mid-tier UK practice Beever and Struthers.

The researchers drew on surveys from more than 200 staff, 50 detailed interviews, targeted skills sessions and four in-depth case studies.

The survey further found that 52% of those surveyed were concerned about how client information is handled and protected when genAI is used.

Around 53% of respondents said they turn to genAI to draft emails or refine written material, while 46% use it to summarise meetings or lengthy documents.

Roughly a third (33%) reported using it to support work in spreadsheets or other software.

The study noted that genAI is most effective for routine, structured work. Accounting tasks that require “nuanced understanding, collaboration or ethical decision-making” still depend on professional human expertise.

ICAS Research and Thought Leadership lead Ewan McCall said: “The message from our research is clear: people, not machines, must remain firmly in the driving seat of professional decision-making.”