AI in accountancy – where is it being used?

As more and more accountants look to make use of new capabilities in artificial intelligence (AI), it’s helpful to share experience of how people have used it to date. So, what areas of the profession are leading adoption, and what problems is AI solving?

AI across the profession

Auditors are at the forefront of AI adoption across the accountancy profession. This builds on the investments in data analytics in recent years which have been made by many larger and mid-tier firms. The FRC’s March 2020 thematic review on audit analytics highlights the extent to which data analysis has become integrated into some areas of audit work by the top six UK firms.

Outside of audit, corporate finance practitioners have been looking at ways in which they could use AI in their work. The ICAEW Corporate Finance Faculty’s report from 2019 on AI in Corporate Advisory outlines a variety of ways in which AI could be used, from deal origination to due diligence to post-deal improvements. However, the report did highlight the gap between potential and reality, with the level of adoption rated by practitioners as around 4/10 compared to 9/10 in its potential to deliver benefits.

Across the profession more broadly, there are a range of examples of early use and experimentation. For example, vendors of cloud accounting software are starting to incorporate AI capabilities in their software to improve automation of processes and deliver better management information to smaller businesses and their accountants. Some large functions also are exploring ways in which they could use AI to improve their operations in a variety of areas.  

Common use cases

So how are accountants in different technical domains using AI and what problems can it help to solve? There are three broad areas where AI is of particular interest:

  • Anomaly and risk detection – for example, AI in audit has focused, in the first instance, on anomaly detection, analysing journals for unusual activity as part of the risk assessment and testing processes
  • Forecasting - given that machine learning is a predictive technology, forecasting activities would be an obvious place to start. Using AI in forecasting can enable the use of more data, often from new sources, leading to more accurate modelling as a result.
  • Natural language processing and generation - AI can be used to analyse text and flag specific risks or issues more effectively than more traditional means of searching documents, whether manually or electronically.

Further resources

The Tech Faculty has a variety of resources on AI, including our free thought leadership reports AI and the future of accountancy and New technologies, ethics and accountability. There is an AI hub page which gathers resources from across ICAEW on AI.

A fuller version of this article is also available to Tech Faculty members.

No Data
  • The ICAEW has put out some great content on this and related topics over the last couple of years… one area I feel we should be better at is differentiating between AI (in the machine learning, neural networks type sense) and plain vanilla computing – the profession still has masses of untapped opportunity in areas such as software interfacing, task automation, data warehousing and analysis even before going near the ‘futurology’ of the AI piece. The IT marketing world is guilty of mis-badging anything and everything as “AI” and we should endeavour as a profession to be more discerning in how we cut through this noisy backdrop. I’ve noticed a few times things like “identifying journal postings made on a weekend” creeping into ICAEW articles as supposed examples of trailblazing firms adopting AI – it is nothing of the sort, but the fact we are able to process more data, more quickly and more cheaply these days is something worth celebrating and promoting in its own right without it always being tagged as “AI”.

No Data