Many auditors were of the view that traditional audit knowledge was still much more important than accumulating IT expertiseProfessor Crawford Spence
10 November 2020
Technology and the brave new world of audit
Crawford Spence, Professor of Accounting and Co-Director of the FinWork Futures Research Centre
How have practitioners and regulators responded to the arrival of the age of machine learning?
Crawford Spence, Professor of Accounting and Co-Director of the Business School’s FinWork Futures Research Centre is studying how audit and the profession of audit will be influenced by the rise of artificial intelligence and machine learning.
He shared his thoughts and some of his early findings with ‘Accounting and Business’ the official news site for the global accountancy body the Association of Chartered Certified Accountants:
Technological developments are reshaping the contours of auditing, drastically increasing the amount of information available to auditors, reconfiguring both staffing models and the size and shape of organisations, and potentially altering the basis on which significant judgments are achieved.
Data analytics, robotic process automation and forays into artificial intelligence (AI) through machine learning are assisting firms in the search for more efficient and reliable audit approaches. These can enable the identification of complex patterns and anomalies in large volumes of data at unprecedented speed.
Alongside efficiency gains, these technologies carry the promise of increasing audit quality and reducing margins of error in the current climate of intense regulatory and public intolerance towards audit failures.
Our research team at FinWork Futures spoke to more than 30 auditors, regulators, software developers and data science experts in and around the profession in order to try and make sense of this brave new world. Below are some of our emerging findings, organised thematically.
Liberation from drudgery
Many respondents viewed the adoption of AI as not reducing the need for human intervention. Rather, AI was being understood as something that would complement human judgmental processes.
‘It’s about complementing human skills,’ said one respondent. ‘The audit profession is going to be highly automated technology-wise, but you’re still going to require a degree of problem-solving and lateral thinking.’
The implication here is that certain processes will be automated, leaving auditors free to do more interpretative and analytical work. In this respect, technology is not seen as a threat, but something that will simultaneously bolster professional judgment and liberate auditors from drudgery.
This all sounds very positive, implying that humans will not necessarily be replaced by machines but will just be given more interesting tasks to do. However, such views are not universal.
Others do see a much more circumscribed role for humans when it comes to making professional judgments, although they equally recognise that we are not at that stage yet because technological developments and, critically, attendant regulatory alignment have been lamentably slow.
Increasing the expectations gap?
Regulatory challenges relate to whether the newly enabled identification of population ‘outliers’, and visualisations now made possible by data analytics from full sets of corporate data, qualify as audit testing within the current auditing standards framework.
This framework requires an auditor to perform a risk assessment to inform the development of a testing strategy. This can then be based either on a statistical random sample of transactions (whereby the sample is traced to the source documentation) or on analytical review procedures that require an auditor to a priori form an independent expectation of a particular account balance in order to ascertain the reasonableness of that figure.
Some consensus appears to be forming among auditors and regulators that this type of technology-enabled analysis should be considered as part of risk assessment as opposed to actual audit testing.
However, if more data analytics ultimately means more risk assessment and less substantive testing (depending on how these notions are interpreted in the context of data analytics), then the risk may actually arise that the so-called ‘expectation gap’ (see ACCA’s recent report, Closing the expectation gap in audit) – between what auditors do and what the public expects of them – widens rather than closes.
Hybridisation – not for the individual
The differing, more analytical and machine-aided nature of the audit implies that new skillsets and forms of expertise are required of auditors. This means that they are becoming more tech savvy, although we didn’t see a huge change in recruitment patterns in firms.
For example, there has been no dramatic shift to employ computer science graduates instead of those from business schools or history departments. Indeed, many auditors were of the view that traditional audit knowledge was still much more important than accumulating IT expertise, even if the latter was growing in significance.
This is not to say that the hybridisation of expertise is not taking place within audit firms. Rather, it is occurring at the organisational rather than individual level. In this regard, we are observing huge growth in data analytics and data science teams at the heart of audit and professional service firms, populated by a mixture of data scientists, accountants and those who are fluent, or at least functional, in both of these languages.
These teams, which can provide support functions to auditors, tax specialists and corporate finance experts among others, are where the major growth areas are likely to be in future years within professional service firms.
Towards a better audit?
There are clearly interesting disruptions taking place in the audit space, and these will continue for some time yet as firms and regulators continue to grapple with the fourth industrial revolution.
What is still unclear is to what extent this all leads to better audits. Many auditors believe that it does, but we have observed some real hesitancy around this issue as well.
One senior partner told us that, with AI and analytics, he can now ‘sleep better at night’. But is this the best measure of audit quality?