The AI Frontier: Promises, Pitfalls, and Professionalism in Technology

Artificial Intelligence (‘AI’) refers to computer systems capable of performing complex tasks that historically only a human could do, such as reasoning, making decisions, or solving problems. 

The reliability (or potential unreliability) of technology has been debated widely in the past few months, predominantly with respect to the Post Office scandal.  But there’s no escaping the fact that technology, in most cases, helps make our lives significantly easier and it seems that technological processes are continuing to advance at an ever-increasing rate with the current “big risk” being Artificial Intelligence (‘AI’).

There has been a significant increase in AI chatter in the tax profession in recent months, largely around its limitations and how it can be used to make better use of an adviser’s time, and the CIOT is running a number of webinars on the topic over the next six months.  Within the chatter, there are of a course a large number of doomsayers who remain convinced that AI will replace humans and eventually take over the world (Goldman Sachs has already suggested that AI could replace 300 million jobs).  Given the well documented limitations (including the case cited below) it’s unlikely that this will happen in my lifetime.

We use Xero for many of our business compliance clients and even ten years ago the software was able to identify a potential bank reconciliation transactions based on the amount, date, and payee (and as time went on, it was able to recognise these more easily), all that’s left to do is check the accuracy and click OK (saving us time and our client’s money).  This is one of the key aspects of AI use in professional services – perceived efficiencies should not come at the expense of quality work (otherwise they’re not efficiencies). 

In a work example, a monthly compliance exercise required the categorisation of 15,000 transactions – the tax team took 33 hours with 95-97% accuracy (anyone would go a little bleary-eyed doing this) whereas the AI model was 97% accurate and took only 5 seconds.  One interesting point is that AI was consistent in its errors making it easier for the human reviewer to identify and correct them.

HMRC has been trying (and arguably failing) for years to use technology to improve service levels – we’ve seen the introduction of MTD for VAT and at some point might see the (continuously delayed) income tax equivalent.  More recently, HMRC customers calling the helpline are being directed to visit the website and use the chat function.  I’ve not had the misfortune to attempt use of that particular service yet, but I’m yet to find one in other industries that actually gives me the results I need, and talking to a person rather than a bot seems impossible.

Anyone who’s had the joy of reading tax legislation knows it’s not always the easiest text to interpret and oftentimes even locating the relevant section can prove troublesome unless you know what obscure title heading has been used.  The same goes for caselaw to a certain extent and this was highlighted in a recent decision in the First Tier Tribunal Felicity Harber [TC9010]:

Mrs Harber had been issued a penalty for failing to notify a charge to Capital Gains Tax (‘CGT’).  She appealed on the basis that she had a reasonable excuse due to her mental health conditions and because it was reasonable for her to be ignorant of the law (you can guess how HMRC feel about that last part).  Mrs Harber represented herself in the tribunal case, she did not seek legal representation or advice, instead relying on the use of AI.  Her submissions included a number of judgements that held support for her case; unfortunately, nine of those cases did not exist (awks).  Mrs Harber’s appeal was unsuccessful.

As with all forms of technology, there are limits to the suitability of its function – could I have asked ChatGPT to write this article for me?  Absolutely, and it would’ve taken a fraction of the time, but aside from the fact that being a member of CIOT means I am required to undertake Continual Professional Development, part of my enjoyment of this profession is undertaking research and ensuring a level of understanding to enable the coherent communication of the findings from that research to non-technical people.

Enjoyed this article then why not sign up to recieve it directly into your inbox

If you wish to discuss anything in this article then please feel free to contact the team here