Artificial intelligence tools can give accountants the freedom to refocus their attention on the highest risks and transform the way they deal with them, but at what cost, asks Nick Martindale.
Barely a day goes by without talk of the impact of AI on society, and 2020 may well be the year when the conversation becomes increasingly focused on the world of work. Until very recently, accountancy firms have been in the so-called play stage of AI, but developments are beginning to accelerate as firms start to create products for specific service lines.
The proportion of organisations using AI in some form rose from 10% in 2016 to 37% in 2019 and that figure is extremely likely to rise further in the coming year, according to Gartner’s 2019 CIO Agenda survey.
One area that is ripe for development with AI tools is in the field of identifying and assessing risk. Its use in assessing risks in audit is perhaps the most obvious area, says Kirstin Gillon, Manager, Technical Faculty, at ICAEW. “AI for auditors is all about looking at transactions and identifying anomalous ones much more effectively and accurately than the traditional ways of doing that, so the human effort can be focused on the areas of highest risk rather than just looking at small samples on a random basis,” she says. “That’s the area where we have seen it used most, and where the use is most advanced.”
Accountants are also using AI to review risks in due diligence around contracts and other pieces of text, Gillon adds. AI tools are used to identify problematic clauses, and could even be deployed in future to assess which businesses are likely to fail based on historic trends. For instance, learning from metrics of failed firms around cash flow, profit margins and exposure to risk and applying these to similarly sized businesses in a comparable sector or location.
AI tools are becoming vital in rethinking risk assessment in a bid to eradicate money laundering and fraud. Traditionally, organisations have sought to identify their customers using only a very limited number of data attributes such as name, address and date of birth. But, says Steve Elliot, Managing Director of LexisNexis Risk Solutions, machine-learning predictive technologies in identification processes could help in the fight against money laundering or fraud.
“Deploying just a few characteristics to authenticate users leads to incomplete identities. The rich, detailed data sets machine-learning models gather enable individuals to be traced, authenticated and risk-assessed fast,” Elliot says. Accountants aren’t redundant yet, even if AI takes on more of the responsibility for assessing transactions or identifying patterns of behaviour.
“The optimal approach across industries involves using machine learning-powered decision-making systems to augment the human workforce while providing transparent, rules-based audits of how judgements were arrived at, allowing human subject matter experts to intervene where necessary,” says James Loft, COO of AI-technology company Rainbird. “To rely only on machine learning is like looking in the rear-view mirror and trying to predict the corners ahead.”
Humans will also be required to regularly sense-check results, and the rules on which they are based, warns Thordis Thorsteins, data scientist at security firm Panaseer. “Some of the key qualities of predictive models and statistical analysis are that they are consistent, scalable and performant,” he points out. “But any such methods are based on a list of assumptions, which can become outdated or reflect the biases of the person who created them. A good solution will be critically reassessed regularly to minimise the chances of its assumptions becoming invalid.” AI does the repetitive work much faster so accountants can delve deeper and focus more time on advising.
“Machine learning and AI do the heavy lifting of calculations – the dull stuff like reconciliations and the tedious work of verifying information,” says Damon Anderson, director of operations at Xero, a New Zealand-based accountancy software company. “This means that finance professionals – especially accountants and bookkeepers – are freed up to focus on the things that really matter, elevating their role to that of coach and adviser.”
As well as having time to focus more on financial risk, AI should also help accountants deliver extra value in other areas. Loft gives the example of teams being better able to tackle new challenges, such as IR35, which could improve efficiency on in-house teams and offer new revenue streams for those in practice. “Teams are at capacity already and the speed of regulatory change and public scrutiny means relating to large changes is hard to achieve and often costly,” he says. “The scale of this means this is not achievable unless these new elements go straight to automated offerings.”
Rick Payne, Technical Manager, Finance Direction, in ICAEW’s Business and Management Faculty believes the impact of AI in generating more capacity for teams can help deliver value, and reduce risk, in other ways, too. “One idea that is particularly interesting is the value of data as an intangible asset, so how data is used, where you can get value out of it and how you can value that,” he says.
“Then there’s the whole area of sustainability and having more time to pay attention to climate change, and the opportunities that might arise from that, as well as the risks. That’s another area accountants can spend more time focusing on if they can free up the time.” That is where accountants will really feel the benefits of AI.
ICAEW’s recent report Digital Transformations in Finance Functions identified challenges organisations face before they can fully take advantage of AI. The study found finance data was generally of good quality but disparate systems led to a lot of manual effort to integrate and reconcile data.
“Most finance functions are grappling with issues around data and trying to overcome legacy systems and non-integration of systems,” ICAEW Technical Faculty Manager Kirstin Gillon points out. “There’s still a lot of paperwork needed in order to get that data insight.” But some businesses also admitted a lack of data standards across the organisation often makes it harder to bring data together.
Some businesses are making concerted efforts to address data governance issues while others are creating data lakes, where information is placed into a separate repository and structured to allow different business functions to make use of it.
In some organisations, finance functions fear the potential for over-reliance on data. “These concerns perhaps reflect the experience that many finance professionals have around data, as well as their natural scepticism and attention to detail,” the report says. “These can be real Strengths of the profession, bringing a realism to discussions and helping business functions to build robust, grounded and value-driven approaches to data.”