Embracing the Future: AI’s Role in Shaping Tax Functions

KT: I’m writing an article on how AI is shaping the tax profession and how we must embrace the future rather than work against it. Can you suggest a good way to open an article?

ChatGPT: “In the ever-evolving landscape of the tax profession, a revolutionary force is sweeping through with unprecedented momentum – Artificial Intelligence (AI). Gone are the days of tedious manual calculations, hours spent sifting through piles of paperwork, and the constant struggle to keep up with ever-changing tax laws. Today, we stand on the cusp of a transformative era where AI is redefining the very essence of how tax professionals operate. The key lies not in resisting change, but in understanding how AI is shaping the tax profession and how we can leverage it to create a brighter future…”

Human back in the room. Promise!

It’s been eye-opening, to say the least, to witness the AI developments of this year alone. The advent of AI technology has immediately been met with controversy, with many fearful of its impact on society, all while we have other groups fully embracing its capabilities.

One of the biggest questions we have about AI is what it means for our jobs and tax. Like countless other professions, the tax sector has been profoundly shaped by relentless advancements in technology, ushering in an era of extraordinary transformations

Technology in Tax: From 2014 to the Present

The concept of a ‘taxologist’ was coined back in 2014 by Thomson Reuters and, at that time, you’d be forgiven for not knowing what exactly a ‘tax technologist’ was. Fast forward 9 years, and tax technologists have become essential parts of any successful tax team, particularly within practice, and more recently in-house.

With the ever-growing influence of generative AI on various sectors, it is evident that the tax industry will not remain untouched.

We’re always commenting on the evolutionary nature of tax – so as soon as the new disruptor  ChatGPT hit the headlines at the end of 2022, we took note. And things got even more interesting in March 2023 when GPT-4 was released and part of its launch was performing a live tax computation. Understandably, this caused excitement amongst some, but equally it has raised concerns with others who are worried about their job being replaced by AI tools. This is hardly surprising when firms like IBM paused their hiring plans to replace 7,800 jobs with AI.

So, how will tax functions adapt to the AI revolution in the coming years? And how will this impact the job market? 

Four ways AI can be used by tax functions 

Over the past few years, digitisation has left an indelible mark on various aspects of taxation, with compliance emerging as one of the most profoundly impacted areas. Despite this transformative shift, compliance continues to pose a substantial time drain on tax teams.

Martin Fiore, Managing Partner at EY recently noted,

“80% of the tax function’s time is spent gathering data and 20% analysing it. AI allows us to almost invert that, creating a massive gain in value.”

Going forward, it is likely that it will minimise burdensome administrative tax work, with its ability to automate tasks including:  

Data collection

Tax professionals frequently invest significant time in rectifying messy data, with SAP formats frequently being described as a formidable challenge. The potential of AI to transform this procedure by extracting data without relying on excessively prescriptive Excel formulas is already showing promise in the realm of tax, even in its current state of development.

Analysis

AI’s ability to comprehend core accounting principles enables it to extract intriguing insights from data. You can feed it excel sheets and ask relevant questions based on the data input which allows for the identification of patterns, anomalies, and potential indicators of tax evasion or fraud. This is definitely an area which is likely to develop quickly, as it has the potential to revolutionise tax enforcement, addressing a potent solution to challenges like the persistent ‘VAT Gap’.

Returns

Research conducted by Tax Systems saw AI being tasked with performing a VAT return, and the results were promising, with a substantial portion of the return being accurately handled, on par with human accuracy for simpler cases. Nevertheless, challenges arise when dealing with more intricate returns involving partial exemptions or overseas taxes. As AI models continue to evolve through current and ongoing training, there’s a strong possibility of marked improvement in this domain. However, achieving human-level accuracy could take some time. 

Advising

With advancements in AI’s natural language processing capabilities, chatbots are now capable of comprehending and responding to complex tax questions. The potential of virtual assistants providing instant access to tax advice has sparked interest, but concerns about their accuracy persist. According to a LinkedIn poll by Borys Ulanenko , opinions vary, with 10% believing AI can currently provide accurate tax advice, 29% expecting it within 5 years, 21% within 10 years, and 40% thinking it may take decades or may never reach human-level accuracy. While AI optimises advisory services, chatbots currently lack the human judgment and critical thinking essential for the tax function, so widespread adoption may take time.

This is just a snapshot of AI’s capabilities within the tax function. But what these 4 uses demonstrate is that tax functions could soon be waving goodbye to basic tasks, in favor of juicer projects. 

Russell Gammon, Chief Solutions Officer at TaxSystems, told Harvey John:

“Having already done some detailed research on the topic, we believe that generative AI has the capacity to remove low-value hours from the compliance process. This isn’t therefore about taking away jobs, more about providing a “co-pilot” for tax professionals to make better use of their time.”

The likes of technologies like ChatGPT hold exciting potential for various tasks within tax functions. But, they also present a plethora of challenges. 

What are the current limitations of AI in tax?

Data Protection 

AI’s major drawback lies in data protection, making most large companies will not be inputting confidential client data into AI tools like ChatGPT anytime soon. Although OpenAI’s policy assures data privacy, sizable enterprises rightfully exercise caution in avoiding client data integration with AI, at least for now. Recent reports of ChatGPT data surfacing on the dark web only reinforce this apprehension. Data security emerges as the most significant obstacle AI must tackle in the coming decade. While alternative language models shielded by internal firewalls do exist, their adoption is hindered by exorbitant costs and the relative novelty of these solutions. Striking a balance between AI’s potential and safeguarding sensitive information remains a crucial challenge in the years to come.

Scalability 

At present, data processing capacity remains constrained. While ChatGPT can manage 8,000 rows, plans are in place to enhance its capabilities to handle a projected scale of 30,000+ rows. Although alternative AI tools exist to address this challenge, they come with a substantial cost, and like many machine learning systems, they are continually evolving and learning. 

Accuracy 

Many people in our tax community have questioned if AI can really replace human accuracy, with many remaining skeptical. Although tax is considered a deterministic discipline, it also requires an element of judgment. But it is important to note, humans make errors too, and incentives like Making Tax Digital have been designed to minimise these. But whether AI could really replicate human accuracy remains uncertain. 

Trust & Regulation

The inherent issue with AI tools lies in their unorthodox coding, making it difficult to understand how they come to conclusions as you could with a computer or calculator. The likes of OpenAI can effectively describe audit trails, referencing them with digital links, vital for VAT functions, remains a limitation. We’re unlikely to see any mass adoption of AI until tax authorities demonstrate their reliability

In terms of regulation, it was only recently the legal sector saw a case thrown out for using fictitious details from AI, making it clear not all information generated by these tools is accurate. So, how will its use be regulated? Again a stance will need to be taken by tax authorities. Over the coming years, we should expect to see a development in controls of how AI can be used which will improve this outlook such as the EU’s AI Act.

Where do we go from here?

As the tax industry evolves, AI’s role in shaping its future becomes clear. Generative AI offers vast possibilities, and can revolutionise the tax profession, freeing professionals to focus on higher-value tasks while AI assists as a “co-pilot.” But there are still creases to iron out. 

Further understanding, adaptation, and collaboration will unlock AI’s true potential, ushering in a new era of tax transformation.

So naturally, we’re now wondering what all this means for the job market, here’s our prediction.