Machine learning and the audit: Rise of the machines?

Machine learning and the audit: Rise of the machines?

Machine learning (sometimes known as cognitive automation) and deep learning (or artificial intelligence) are two of the three forms of what is sometimes termed ‘digital labour’, with the third being robotic process automation (RPA). On a spectrum of technological advancement, RPA is the most basic, machine learning is more sophisticated and deep learning is the most sophisticated of all.

Illustration of cogs in a human head

Where RPA uses technology to automate a process such as collecting data, machine learning uses algorithms to analyse data and make correlations and predictions (with human oversight). It is a more ‘intelligent’ form of technology than RPA.

Deep learning is where the technology appears truly intelligent as the machine may learn from its own experience, teach itself how to perform a task or analyse it, and so improve its own performance or effectiveness. It is as though the technology has a brain.

Machine learning and the audit

So, what are the implications of machine learning and deep learning for the audit?

Along with RPA, machine learning is arguably the biggest new technology already at play in the audit. On top of RPA processes, we are also applying machine-learning techniques where, through complex algorithms, the technology can scan information, model it against thousands of assumptions drawn from external scenarios and highlight risks and insights. This predictive analytics is a step towards deep learning where, in the future, the application will be able to ‘think’ for itself, learn from the results and run more scenarios and tests accordingly.

Machine learning is a vital step beyond robotics because the technology can capture data and identify correlations and patterns. It is also more intelligent than robotics — for example, it can locate specific line items by currency symbol and/or other keywords even if the placement varies from invoice to invoice — something robotics alone is not able to achieve. The technology can therefore be used to scan huge volumes of information. For example, on various valuation matters, we are developing applications to read bank loan agreements, leasing contracts, and other documents in order to find specific data and identify subjective areas.

Our smart audit platform, KPMG Clara, is helping us to present our findings in powerful visualisation capabilities to organisations called a ‘visual ledger’. This enhances our understanding of transaction flow and allows us to ask more precise questions on those transactions that stand out from the others.

Deep learning

Deep learning — full-fledged artificial intelligence where a machine continuously integrates new information, draws conclusions and absorbs the learnings to enhance its cognitive abilities — is seen as one of the greatest potential prizes of emerging technology.

While full utilisation of deep learning is in the near future, KPMG is well positioned for it through KPMG Clara, which has been developed specifically to enable new technologies to be integrated, and the platform is fully scalable.

There are issues that will need to be resolved. For example, it will be critical to show that the algorithms and technology behind deep learning are valid and robust in order to understand and document the outcomes derived and not to be lead to ‘false positives’.

Regulators are actively working to keep pace with these emerging technologies. In the US, the American Institute of Certified Professional Accountants (AICPA) has published a guide to data analytics in the audit.

What is the impact on audit quality?

There is a clear consensus across KPMG’s network of audit professionals that the array of new technologies being developed will help to increase audit quality. The ability to analyse 100% of datasets rather than sampling provides obvious benefits. The technology will enable auditors to focus their efforts on the outliers and anomalies, devote greater time to areas of higher risk and have meaningful conversations — resulting in a better quality audit.

The power of robotics, machine learning, natural language processing (NLP) and, in time, deep learning will mean that an audit may become deeper and further-reaching than ever before, based on increasingly granular and sophisticated analysis of data. Combined with the integration of our applications into KPMG Clara, its flexible deployment in the cloud, and with such capabilities as Dynamic Risk Assessment, we are well placed to develop ever more powerful capabilities. These capabilities will also help drive consistency across the network, helping us to enhance service to cross-border businesses.

The fundamentals of an audit will not change as the need for human judgment and professional scepticism will always be necessary. The real-use case for new technologies is that they will enable us to obtain — more easily, quickly, accurately and extensively than ever before — the corroborating evidence that is needed in an audit.

KPMG Australia acknowledges the Traditional Custodians of the land on which we operate, live and gather as employees, and recognise their continuing connection to land, water and community. We pay respect to Elders past, present and emerging.

©2022 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.

Liability limited by a scheme approved under Professional Standards Legislation.

For more detail about the structure of the KPMG global organisation please visit

Connect with us

Save, Curate and Share

Save what resonates, curate a library of information, and share content with your network of contacts.