Why human judgement still matters in the age of Gen AI
More than two-thirds (72%) of accounting professionals fear that generative artificial intelligence (Gen AI) could produce errors or reach incorrect decisions, according to a year-long research study commissioned by the Institute of Chartered Accountants of Scotland (ICAS) and involving academics from AMBS.
The findings come from a major new report ‘Generative AI and Professional Judgement in Accounting’, published as part of ICAS’ Shaping the Profession programme. The research project outlines one of the most comprehensive examinations to date of how Gen AI is influencing accountancy and professional judgement, highlighting the critical importance of maintaining human oversight as the technology takes on a more routine role in daily work.
“The research highlights that Gen AI should not be treated as a uniform, firm-wide capability. Its risks and benefits depend on the type of work involved and where this distinction is lost it becomes harder to make responsibilities visible.”
Real-world insight
Conducted by research partners from AMBS and Aston Business School, the study provides rare real-world insight into how Gen AI is being adopted at the mid-tier UK accounting firm Beever and Struthers (now part of Menzies LLP).
Drawing on firm-wide surveys of more than 200 accounting professionals, 50 in-depth interviews, targeted skills development activities, and four detailed case studies, the project explores accountants’ use of the technology in day-to-day practice and how it is shaping their professional judgement.
The AMBS team, which worked on the project, included Yasmine Chahed, Sung Hwan Chai, Brian Nicholson and Leonid Sokolovskyy, alongside George Salijeni from Aston University. Menzies LLP colleagues included Caroline Monk, John Toon (now at HLB Network) and Bec Hesketh.
“This report shows that while generative AI can make accounting work faster and more efficient, professional judgement, ethical oversight and human accountability remain absolutely essential.”
Key findings
- Gen AI is boosting productivity: Nearly three-quarters of the accounting professionals surveyed (74%) say the technology speeds up the tasks they perform, with many highlighting quicker data extraction, document analysis and information cross-referencing. Yet fewer than a quarter (22%) say this has translated into a substantial increase in their overall work rate.
- Accountants are cautious about accuracy: More than two-thirds of respondents (72%) worry that Gen AI could produce errors or incorrect decisions, highlighting the need for careful review and strong oversight of outputs.
- Beware of the ethical and oversight risks: More than half of respondents (52%) raised worries about client data privacy and confidentiality. The research stresses that professionals must also stay alert to hidden cognitive biases that could skew their judgement.
- Routine tasks remain the most common use of Gen AI: More than half of respondents (53%) turn to Gen AI to write emails and edit text, while 46% use it to summarise meetings or documents. Around a third (33%) rely on it to support spreadsheet or other software-based tasks.
- Human judgement is essential: Gen AI performs best on routine, structured tasks but accounting work that demands nuanced understanding, collaboration or ethical decision-making continues to rely on professional human judgement. This helps explain why only a fifth of respondents (21%) said they were concerned Gen AI might affect their job security.
“The use of generative AI in accounting and auditing extends well beyond simple summarisation of documents and standards. We observed professionals using it as an interactive sparring partner to organise information, explore alternative lines of reasoning, and support all stages of the judgement process.”
The report urges a clear, accountable approach to the responsible adoption of AI across the accounting profession and beyond. It calls on organisations to place ethics and human judgement at the heart of AI use, introduce safeguards to limit bias and over-reliance on AI outputs, and set clear frameworks for when and how the technology should be used.
The study also emphasises the need for strong oversight, clear accountability for decisions, and proactive steps to ensure AI does not widen skills gaps or restrict access to the profession.
“This report demonstrates that while Gen AI can significantly accelerate the analysis of accounting guidance, it remains an assistant, not an author. The efficiency it provides is only valuable if it is anchored by the unwavering accountability of the human professional.”
Judgement and accountability
ICAS Research and Thought Leadership lead Ewan McCall said the findings offer both an opportunity and a warning to the profession that while Gen AI may be accelerating routine work, it can't replace the judgement and accountability that underpins public trust in accountants.
He said: “The message from our research is clear: people, not machines, must remain firmly in the driving seat of professional decision-making. While Gen AI can help accountants work faster and handle many routine tasks more efficiently, the technology is only as effective as the information, supervision, and oversight behind it. Weak inputs lead to weak outputs, which over time erode the trust that underpins accounting work and wider society.
“Backed up by insight from research like this, the accountancy profession and every sector using Gen AI can rise to the challenge of making sure the technology is used critically and ethically to strengthen, not stifle, professional judgement.”