Artificial intelligence (AI) is rapidly transforming the world of work, and the tax industry is no exception.

However, against a backdrop of growing use of AI in tax to streamline processes and increase efficiency, concerns are rife that AI could lead to job losses. Goldman Sachs has already suggested that AI has the potential to replace 300 million jobs.

There was some excitement (and trepidation) when, as part of the launch of ChatGPT’s latest version (GPT-4), the AI tool performed a live tax computation, no doubt fuelling concerns among tax practitioners about what this means for their day-to-day jobs.

Ian Pay, Head of Data Analytics and Tech at ICAEW, believes that job security concerns among tax experts should not be overplayed. “Certainly it feels like areas where there is a clear set of rules to follow, and clear, concise inputs to which those rules can be applied, are ripe for AI to step in,” he says.

“But while tax is predominantly rule-based, there are ambiguities, nuances and a huge array of factors based on individual context. This is where tax practitioners can still offer services far above the AI tools, to actually help clients pay the right amount of tax based on their circumstances – no more, no less,” Pay adds.

Mazars’ use of AI

Catherine Hall, Head of Tax at Mazars, says there’s great potential for AI to improve efficiency through swift data collection, thereby enabling tax professionals to focus their effort on using that information for decision making.

“AI can add an awful lot from a business perspective, but my view is that you will always need that human interface,” she says. “I don’t think we’ll become obsolete, and I think it will remain an attractive profession.”

Hall says tax is an area where a lot of intervention is required and a wide spectrum of skills is needed to get the job done. But the skills of tax practitioners must constantly evolve to adapt to the increasing complexity of the global tax landscape, and incorporating AI into their way of working is a natural evolution of that.

Mazars is already using AI to streamline the tax services it offers. Specifically, AI underpins a tool to look at Public Sector Accounting Standards (PSAs) and sift through vast volumes of data to check that the categorisation is correct. The firm is also exploring other areas where AI can be used for verification against huge data sets.

There are other specific accounting operations AI can help with, including collating data for tax returns, particularly for those operating internationally, and flagging up any unusual patterns. “Collecting data through AI makes it a lot easier to spot anomalies or trends in case anything looks unusual. You could see a trend in declining effective tax rate, for example,” Hall says.

The tax authorities’ perspective

Of course, tools and technologies available to tax practitioners are also available to the tax authorities. Hall warns that they could potentially use AI to spot anomalies and identify where taxpayers are paying less tax than their peers, and pick up more quickly on areas they want to focus on for an investigation.

Caroline Miskin, Senior Technical Manager of Digital Taxation at ICAEW, agrees that use of AI by the tax authorities and regulators could have a dramatic impact on efficiency. “AI has huge potential where transactions need to be processed on a very large scale (whether by taxpayers, advisers or tax authorities) or complex legislation and regulations (perhaps from several jurisdictions) need to be sifted through to identify the potentially relevant rules.”

However, Miskin says its use also raises important questions about who is responsible if AI gives the wrong answer. “In my view the answer provided by AI needs to be explainable, with the end user being able to understand the algorithms and therefore be in a position to take ultimate responsibility,” Miskin says.

“One of the main limitations may prove to be AI’s need for data to be held in a clean, structured way; too often data is captured inaccurately, inconsistently and in different formats. This has certainly been a factor in delaying HMRC’s use of robotics to process paper forms.”

• For more on this subject, listen to ICAEW’s recent podcast, How is HMRC digitalising taxation?, where the Institute discusses the ways in which HMRC is using digital technology to influence tax calculation and collection.