This article was first published in the January 2020 International edition of 
Accounting and Business magazine.

Advances in artificial intelligence (AI) are continuing to rip up the rulebook for all industries and professions. In this ‘fourth industrial revolution’, tax is no exception to the application of technology. A number of studies have examined how AI is likely to impact work in one of the most complex of financial areas.

PwC is the latest to add to the collection of reports in this sector. How tax is leveraging AI in 2019 sets out what distinguishes AI from other emerging technologies, and why it will have such a disruptive effect on tax processes.

Put simply, AI is the ability of a computer or machine to perceive its environment and perform tasks that normally require human intelligence. In theory, it allows the machine to sense, think and act in ways that outperform human capability. This aspect is important – it is not a case of matching human performance, but improving on it and ultimately replacing it. At least, that’s the theory, although many finance professionals maintain that a human element will always be required.

PwC argues that the tax function, like so many other aspects of the finance operation, needs to be nimble and accurate, able to provide information that the enterprise needs for business decisions. But there is another requirement pillar that stands alongside this – compliance and the increasingly complex rules and calculations that apply to tax reporting with an ever increasing international dimension.

So, on the one hand, tax functions face the issues of global tax reform, pressure to reduce costs, the need for better and faster data, the adoption of emerging technologies by the enterprise, and the demand for new skills. On the other, digital is infusing everything, effective predictive and calculation models are required, financial and reputational risk must be managed, adaptability is all, and new trade and customs considerations are arising. Technological disruption links all these elements.

PwC splits AI’s capabilities into three areas: sensing, thinking and acting. ‘Sensing’ includes natural language processing (for example, the scanning and conversion into text of tax notices), and machine learning, where text is understood and data extracted.

The machine learning abilities of AI are also seen in ‘thinking’, where statistical learning techniques are applied to automatically identify patterns in the data.

‘Acting’ will include the simplest form of AI, robotic process automation (RPA), where machines mimic human actions. It also includes cognitive automation, which makes inferences based on information found in unstructured data, as well as the ability to answer deep questions by searching large databases.

PwC suggests a responsible framework for implementing AI within the tax function, which covers strategy, design, development and operation. The strategy needs to align with the business and the tax function; the design should build in transparency, embed controls and be user-friendly; development must involve strong programme management; and the operation should eliminate unintentional bias, and prevent malicious attacks, while preparing for any potential systemic risks.

The opportunities for the tax function are clear. As Martin Fiore, managing partner in EY’s eastern US region tax practice, wrote recently in a blog: ‘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.’

His colleague Channing Flynn, EY’s tax technology leader, adds: ‘As powerful as AI is today, it is still in its infancy.’

Flynn believes the coming decade will see huge leaps in the way in which AI is used and developed. For instance, he believes that AI will allow unprecedented reach and insight into data, transforming how transfer pricing comparables are done.

‘Currently, transfer pricing is more art than science and a major challenge for both taxpayers and governments alike,’ he says. ‘AI will also radically change controversy and planning – which are time-consuming and complex areas. AI will allow companies to manage risk and do predictive analyses far more quickly and powerfully than is possible today. Companies will be able to gain rapid insight into case law and procedural challenges and all that’s available in the public domain to help inform a tax strategy or to update tax compliance procedures for new law.’

Of course, there are the well-rehearsed fears that AI will replace people, causing job losses in the process. The positive spin is that AI will enable tax professionals to work better, smarter and faster, which will mean more time to carry out value-adding activities, from managing risk to partnering with the business on strategic projects. It may even encourage recruitment into the profession. As Fiore says: ‘The higher-level, higher-tech work that AI allows will encourage the best and brightest to enter the profession; it will help them learn faster and it will provide greater career satisfaction.’

In its study Artificial intelligence: entering the world of tax, Deloitte points out that while some tax professionals fear AI will erode their perceived added value, the reality is that AI deployments are more likely to allow them to focus on the more rewarding intractable problems and leave the lower-level tasks to the machines.

Philip Smith, journalist