Risk & Economy » Audit » Big Four firms raising audit bar with AI

Artificial Intelligence (AI) is beginning to revolutionise the way companies are audited. But how far are the leading accountancy firms in developing this technology and what do they see as the benefits to their clients?

The argument being advanced by the leading accountancy firms is that AI can improve the quality of audit work by carrying out tasks faster more accurately, than humans.

They say AI has the potential to assess huge volumes of data and generate new types of insights, recognise complex patterns and detect anomalies, all powerful elements in the audit process.

Gilly Lord, head of audit strategy and transformation at PricewaterhouseCoopers (PwC) says there are a whole host of ways in which AI has begun to change the way the firm audits. One example is a tool PwC has developed called GL.ai which identifies anomalies in a business’ general ledger

“We’ve trained the tool using over 1 billion journals and now, using advanced algorithms, it can analyse billions of data points in milliseconds and apply judgment to spot items of potential error or fraud.

“These algorithms see what humans can’t – analysing every transaction and identifying areas for additional investigation – meaning that auditors can focus their efforts where the risks lie,” says Lord.

“Tools like these are helping auditors to work faster and improve accuracy. The technology performs the high volume tasks involved in the audit, meaning the auditors can concentrate on the tricky, subjective areas which require experience and judgement,” says Lord.

“This not only offers them a more fulfilling role, but also means they can present deeper insights back to a business and its key stakeholders,” she adds.

Katie Canell, UK audit innovation lead at Deloitte says it’s important to make the distinction between AI and cognitive. “In an artificial intelligence system, the system will tell the auditor which course of action to take based on its analysis.

“In cognitive computing, the system provides information to help the accountant or auditor decide and apply human experience and judgment,” she says.

Whilst AI capability may be less embedded into the standard end to end audit approach than – say – analytics at the current time, instances of cognitive capability are very much in existence, says Canell.

Deloitte has invested in cognitive capabilities that enable the firm to use natural language processing to read documents such as contracts, in addition to investing in natural language generation capabilities to automate the production of written documents.

“In addition, we have developed a voice recognition analytical platform which analyses emotion and behaviour through voice interactions,” says Canell.

 Investing in intelligence

Training the predictive models used in both cognitive and AI software is often an extensive and complex process and requires an upfront time investment in order to create an effective solution for the audit.

However, as with many complex technologies, starting small, remaining agile in development, and constantly refining the product will create a successful longer-term solution, says Canell.

She believes standardisation of data and processes is a prerequisite for use of cognitive technologies – as indeed it is for analytic analysis, or application of robotics process automation.

A certain amount of data housekeeping needs to occur before the auditors can benefit from these digital technologies in their fullest sense. In addition, cognitive and AI technology is still evolving so we must carefully select the most appropriate opportunities where these technologies can be used at this time to achieve the high quality output required,” says Canell.

“The current reality is a world in which examples of cognitive capability combine with the application of human judgement.  This in and of itself creates a powerful and effective transformation of the audit process and is in many ways a stepping stone on the journey of experimentation around more extensive AI enablement where the balance of artificial intelligence vs human intelligence is likely to shift,” she says.

Canell says the benefits are considerable, best illustrated in the area of natural language processing. “Using cognitive technology to read contracts and invoices, auditors can efficiently process, highlight and extract key information across a complete data set, rapidly understanding of the entirely complex contract and identify matters of interest,” she says.

Such advancements have benefits on many levels, not least ensuring consistency of approach across large data sets and giving the ability to effectively and efficiently review whole populations for items of interest.

It helps to focus effort on key items of interest and insight whilst the repetitive, or low judgement, tasks are automated, says Canell. “From a client perspective there is a benefit from rich insight across the entire data set, such as understanding deviations from standard clauses across lengthy, complex contracts,” she says.

“When considering the application of cognitive solutions it is critical that the primary driver is the business challenge and then the technology application should be focused to deliver against that.

“If the starting point is a piece of exciting tech, you may fall into the trap of using technology for technologies sake, and not solving a problem which is critical to the business or gives appropriate return on investment – be that efficiency, quality or effectiveness,” says Canell.

The new paradigm

Hywel Ball, EY’s UK head of audit, says new technology such as data analytics, AI and robotic process automation is changing both what and how audits are undertaken. “It is enabling us to search, sift and sort through large quantities of data – from company reports to social media – these tools are helping auditors to identify potential areas of risk and to understand a company’s performance at a more granular level,” he says.

They are also providing insights into areas that were once thought to be impossible to measure, such as culture. “These insights can be really valuable to clients. What was once called the auditor’s ‘nose’ – or gut instinct – is increasingly being automated by advanced data analytics,” he says.

“AI can automate a lot of knowledge gathering, and will help free up our people to focus on higher-value strategic work and provide more fulfilling responsibilities. These developments will in turn create opportunities to develop new roles and positions for the future, while the skills we need will also continue to evolve.

“In five years’ time, will we be recruiting for ‘data assurance associates’ or ‘blockchain auditors’, rather than just general audit graduates following the standard institute curriculum – who knows?” says Ball.

“The challenge for the profession will be to keep pace with these developments. However, at a time when there is a widening trust gap between business and society, fake news is increasing and there is an explosion of big data, I think the use of new technology will continue to have a vital role to play,” he says.