How is Al changing the tax function? Santhie Goundar investigates the changes firms are seeing; VAT case study
It’s well documented how tax authorities AI is being used to close tax Gaps. But what about VAT? By Santhie Goundar for Financial Accountant.
Of all the taxes, Al is meant for VAT, says Richard Asquith, CEO of VATCalc. However, he notes that while the first wave of AI around machine learning has been popular with tax authorities globally, the tax profession itself “has been slow to follow suit”. Machine learning, he explains, “covers the use of huge amounts of data and algorithms to imitate the way that humans learn, gradually improving its accuracy of basic processes.”
Asquith says there are two reasons Al suits VAT: “Firstly, Al craves data to analyse for trends and best practice to find the right answers – VAT is about millions of daily transactions in the economy, so it’s a perfect data-rich environment Al can learn from.” Secondly, VAT is based on stable tax law and cases: “There are always grey areas on interpretation; but there are fixed rules Al can ingest and apply against these millions/billions of transactions to identify errors or fraud,” he explains.
VATCalc recently released its AI VAT Advisor tool, which uses AI. Asquith sees it as an “efficiency tool” for external advisers – giving them support to prepare a first draft of potential tax advice for their clients.
A note of caution
While Al and automation offers opportunities to do things better, there is an element of caution required, says lan Bowden. “Tax professionals need to question and validate the outputs from Al improvements to their systems,” he says. “Getting carried away with Al without the right technical review and control environment could easily cause tax outputs to be wrong.” Rachel Milloy agrees: “Tax professionals should be cautious of the hype associated with generative Al and exercise judgement over where to commit effort to areas that are likely to benefit the most.”
Asquith believes tax practitioners “should look beyond ChatGPT and the internet – that’s not the threat or opportunity”. It may take years for taxpayers to know and trust chatbots for financially sensitive tax issues, he adds: “The tax profession – like law – is known for well-regulated, educated and trained providers; whereas who is overseeing the bots? Who is responsible for any error by a bot?”
For now, Asquith believes trained and experienced professionals are irreplaceable, but adds: “Al won’t take your job – but someone who embraces Al will.”
