How generative AI is transforming in-house tax practice

The tax profession has been quick to identify the potential of AI and start adopting Gen AI solutions into tax workflows.

A May 2024 survey of UK in-house tax practitioners by Tolley found roughly one-tenth (9%) are using generative AI for their work on a daily basis, while more than a third (36%) are using generative AI at least once a month.

One-tenth of in-house tax practitioners use generative AI every day.

A further 31% of in-house tax practitioners said they are currently not using generative AI but have plans to do so. Only a third (32%) said they have no plans to use generative AI.

Given the pace of change and how much we expect the critical reasoning abilities of generative AI to increase, the tax sector will look very different in a year's time, says Hayley McKelvey, Tax and Legal – Digital Innovation Leader at big four accountancy firm, Deloitte.

"We are only at the foothills of what GenAI can do and how much of the tax professional’s compliance and advisery work it can consume," she says.

Bivek Sharma, Chief Technology Officer & Head of Alliances for Tax, Legal and People business at PwC, says AI has changed the rules of the game, with an unprecedented level of transformation in tax .

"Gone are the times where you have to spend hours, days, or even months reviewing thousands of technical documents."

Whilst the technology doesn't have the emotional intelligence of a human, Sharma says it is highly effective in complementing their people approach.

"The combination of human touch and technology allows us to accelerate deliverables, react quicker to client queries and obtain insights much more effectively than before. "

Helen Whiteman, the CEO of the Chartered Institute of Taxation, says generative AI will only get better.

"We’ve seen rapid development over the past 18 months and clearly AI offers many benefits."

Ian Hayes, the President of CFE Tax Advisers Europe, says for technophiles, generative AI is an amazing new tool, but for technophobes, it is a disaster.

"The speed with which digital products, software programmes, adaptations and updates are being made available is akin to a new dot.com bubble. Everything is now imbued with the magic ingredient that is AI."

While AI might not have an immediate impact on the short and medium term demand for tax services, Hayes says in-house advisers would be wise to start future-proofing their businesses.

"The tax function must be fit for purpose and embrace the inexorable advance of digital by retraining, recruiting and maintaining an agile stance for their businesses."

In-house tax practitioners' top AI priorities are researching tax matters (95%) and drafting tax documents (84%).

We also asked our respondents how they plan to use generative AI in the immediate future. The biggest priorities were researching matters (95%) and drafting documents (84%). Real-time comparisons of law across jurisdictions was also listed as a top priority by 76% of respondents.

AI will help in-house teams to find patterns, issues or common themes that they didn’t know existed, says Adrian Henderson, the Director of Technology and Transformation Services at one of the UK's biggest accountancy firms, Evelyn Partners.

"Generative AI can spot patterns that humans cannot – this helps us create new business insights for our clients," he says.

In-house teams are significantly more likely to have created an AI policy (24% vs. 8%), launched an AI-powered product for internal use (18% vs. 10%), and hired AI experts (11% vs. 5%) than private practice practitioners.

Almost half of in-house teams have made a change to their day-to-day operations as a result of AI.

In-house teams also expressed high expectations for their firms to use generative AI, with half (53%) expecting their firms to use the technology to increase efficiency.

One advantage to in-house teams, suggests Jane MacKay, Corporate Tax Partner at Crowe, is the analysis of large company-wide datasets, such as Import and Customs’ Duty classifications.

"Generative AI means we can quickly identify errors or trends typically resulting in Customs Duty savings for our clients."

Mackay also expressed an interest in running broader datasets through generative AI to highlight risks and future growth opportunities for clients.

Sign up to the  Tolley+ AI™ Insider Programme  to get updates, early feature access and breaking news on the latest AI developments in tax. 

Despite their eagerness, in-house tax teams still have concerns over the ethical implications and risks of using public access AI tools.

Without training, oversight and grounding in robust datasets, generative AI technology can lead to inaccurate answers. The tools also require robust security to ensure confidential data is kept secure.

Jonathan Scriven, Director of Tax Markets at Tolley, says accuracy of information is critical to internal tax departments.

"There's a huge fear of providing the wrong advice to the business, which is particularly heightened by issues with public access generative AI."

76% of tax teams voiced concerns about the reliability and security of public access generative AI tools.

Whiteman from the Chartered Institute of Taxation says data security and privacy pose significant risks to tax practitioners.

"The lack of transparency around how AI systems operate can leave tax practitioners and their clients in the dark."

People who understand the basis of these generative models, and can explain their outputs, need to be included in the loop, she says.

"The big risk here is that the professional services market diverges from the knowledge, skills and abilities on offer from practitioners."

Kay Firth-Butterfield, executive director of the Centre for Trustworthy Technology, a World Economic Forum Centre for the Fourth Industrial Revolution, says these systems are only as good as the data in them.

Generative AI will become a powerful and increasingly robust tool for in-house teams, says Laurence Kiddle, Partner, Head of Technology and Transformation Services at Evelyn Partners.

Despite his enthusiasm, Kiddle warns of the challenge of keeping risk and review models in step with new technology, and ensuring the right level of human review is in place.

"Over-reliance on AI to deal with queries without proper consideration of the issue puts tax teams at high risk of presenting incorrect responses, which could ultimately lead to criminal or civil convictions."

Overreliance on public access generative AI platforms can be particularly problematic for tax research. MacKay from Crowe says the current AI tools for the tax market are limited by the quality of information, context, and appropriateness of research terms.

"The quality of information isn’t currently reliable enough, which can lead to ludicrous or inconsistent results if you don’t already know what answer you are expecting."

This also comes down to the proficiency of the tax professionals, she says, who will need to morph into prompt engineers. "Whatever the quality of the information, if you ask the wrong question, you’ll get the wrong answer."

Hayes from CFE Tax Advisers Europe says the possibilities of generative AI are exciting, but the tax world must have competent risk evaluations in place.

"It is difficult to overstate the importance of having a clean, extensive training database, which is relevant to the business and the subject of the research, and is able to demonstrate issues of bias in the data."

We are still at a very early stage of this technology and have much still to learn, he says. "This is not Artificial General Intelligence and we are not, at this stage, preparing to pass over the controls to a machine."

We asked respondents how confident they would be using a generative AI tool that was grounded on tax content sources, with linked citations to the verifiable authority used to generate the response.

Four-fifths (83%) of tax teams said they would be somewhat or completely confident using AI-powered tools grounded in tax content sources.

Four-fifths of tax departments would be more confident using AI tools grounded in tax content sources.

To work frictionlessly with AI-powered tools, practitioners will need to feel comfortable using and critiquing generated outputs, says McKelvey from Deloitte.

"Tax professionals will need to be far more fluent in data and data interrogation techniques than they are today, even in a world where AI and GenAI is doing most of the heavy lifting."

In-house tax professionals also expressed a strong demand to be informed if the external advisers who they are working with had used AI. Four-fifths (83%) of in-house teams expect to be informed, whilst only half (50%) of those in private practice think that their clients want to be informed.

Tolley is part of RELX, a global provider of information-based analytics and decision tools for professional and business customers, which has clearly-established responsible AI principles in place.

"As part of our responsible approach to developing AI solutions, we always consider the real-world impact of the solution," says Scriven. "We proactively look to prevent the creation or reinforcement of bias by ensuring we can always explain how and why our systems work in the way they do, build in human oversight, and respect and champion privacy and data governance."

Sign up to the  Tolley+ AI™ Insider Programme  to get updates, early feature access and breaking news on the latest AI developments in tax. 

One potentially problematic point of conflict for in-house and private practice teams is pricing. Some believe that the firms investing in generative AI-powered products or tools should be the ones who benefit from the time- and cost- savings. Others believe the benefits should be passed on to the client.

Our survey found almost half (45%) of in-house teams expect bills to be reduced as a result of generative AI. Those in private practice were less likely to believe their clients would expect a reduction in cost, with 35% agreeing.

45% of in-house teams expect bills to be reduced as a result of generative AI.

Scriven from Tolley says generative AI will ultimately mean some tax tasks are undertaken much faster than the current delivery model, which could have an impact on pricing.

"The impact of this on price will depend on many different factors – the importance and number of those tasks in the overall workflow, the pricing mechanism at play, and the commercial relationship between the tax department and their adviser."

There was an even starker contrast in opinion when it came to the topic of restructuring pricing models in general.

Only a quarter (28%) of medium and large practices said they will make changes to their billing as a result of AI, despite 43% believing it will reduce overall costs for the firm and 39% believing it will increase hours billed.

More than half (57%) of in-house teams, however, said they expect practices to make changes to billing as a result of generative AI.

At larger organisations, efficiency gains and increased effectiveness will likely come down to collaboration between tax professionals and tech teams.

Automating the routine, entry-level tasks, such as data analysis and meeting summaries, will mean the skills required of a tax practitioner will change, says Adrian Hextall, Director of Technology and Transformation Services at Evelyn Partners.

"Many were worried when Excel and Lotus-123 replaced the original paper spreadsheets… and the profession adapted and survived. It will do so again but perhaps the role of the tax practitioner will move closer to a tax law-led data scientist."

With increased automation, Jane MacKay, Corporate Tax Partner at Crowe says one concern is how the next generation of tax professionals will develop the skills they need to become well-rounded tax practitioners.

"We need to ensure they acquire the right skills to perform different types of work as the simpler tasks become automated."

Sharma from PwC says curating diverse teams of creative and technical specialists has been useful in providing a broader perspective of the impact of AI.

"Pairing deep technical experts alongside tax subject matter specialists often results in more innovative solutions to complex problems."

In-house teams and private practices have a lot to consider over the months ahead – and pricing will be one of them. Yet the opportunity for tax professionals to improve their reach and break new ground is high. As a result, we are seeing real collaboration between tax departments and technology and innovation teams.