I know we all like to say that doing AI means realistically avoiding doing the more boring, repetitive stuff – and there’s nothing that seems more enticing than automating the tax action really. But if you’ve ever worked in tax, you will know that this is not the end of the story.
When I was an articleship, I spent many weeks perched in client files – roving through ledgers line by line, sifting through vouchers, cross checking TDS balances and making sure each and every number rang true right down to the trial balance. There were days I checked the same schedule three times and I found that a simple floating decimal point was the cause of the mismatch. It was meticulous work – often mind-numbing work – but one provision of a tax call wrongly could easily be a notice or a reputational mess. It was at this moment that I realized tax was not difficult because of mathematics but difficult because of meaning. The key is to get to the why that lies behind all of these numbers.
So I was both impressed and interested when I read about H&R Block’s AI Tax Assist, an AI-based tax service that allows users to ask tax questions in natural language and get abbreviated fact-based answers plus talking points to address the topic. It’s that sort of thing that makes us all wish that someone had invented that thing back in the old days. But it also got me thinking, would I have relied on an AI using a black box algorithm for my taxes? Probably not. Because as perfect as AI can read over every part of the tax code it still can’t read intent.
That’s why I consider Generative AI most suitable in places where it is needed, not for replacement, but as a digital co-pilot. It is excellent at doing the heavy lifting such as raw data extraction, cleansing that data, mapping data to trial balances, identifying anomalies or even drafting up summaries. It is the type of work that needs to take places for hours the old fashioned way, but seldom requires complex judgment. By allowing AI to do all this, professionals are finally able to start dedicating more time to thinking – analyzing, interpreting, and advising.
But responsibility on part of man, also increases. The final advice will still need validation of the AI’s output, application of context and ensuring the generated advice isn’t merely compliant but commercially reasonable. Because there needs to be recognition that tax is not just a technical exercise – it is a judgment call based on law, intention and business reality.
I think, AI is providing us with something we were deprived of gradually by the profession – time to think. Time to ask “why,” not just “what.” Time to get to interpretation rather than transcription.
The future of tax is for those who can bring together three things – technical expertise, digital capability and ethical character. People who are able to translate what AI outputs into what businesses need.
So yes, AI will replace parts of the tax work. But rather than replace us, it may simply remind us why we got into the profession, not to crunch numbers, but to understand what they mean.
Gen AI: A Tax Advisor?
7
October
2025
