Singapore’s Big Four auditors tap AI to streamline audit processes, enhance efficiency

EVEN as artificial intelligence (AI) adoption upends industries worldwide, Singapore’s biggest professional services firms have no plans to reduce hiring in their audit divisions.

In November, The Wall Street Journal reported that KPMG had laid off hundreds of employees in its US auditing division. While studies have shown that accountants face a high risk of job displacement due to AI, KPMG denied that the layoffs were tied to AI adoption.

Earlier this year, PwC in the US also announced plans to lay off 1,800 employees as part of a business streamlining effort.

KPMG in Singapore said it remains focused on growth. “We do not have any plans to conduct layoffs in our workforce,” a spokesperson told The Business Times.

“On the contrary, we are focused on expanding our team by bringing in top talent to further strengthen our capabilities and support our clients, people and communities in Singapore.”

The other Big Four accounting firms – Deloitte, EY and PwC – similarly confirmed no plans to downsize their audit teams locally.

Instead, they aim to leverage digital technologies – particularly generative AI tools – to automate routine tasks and enable auditors to focus on higher-value work.

The intention is not to replace employees with technology but to enhance the efficiency of audit processes, noted Yang Chi Chih, audit leader at Deloitte Singapore.

This, he added, frees up the “time for our auditors to flex their capabilities to focus on delivering higher value work, such as analysis and insights, as part of the auditing process”.

AI use cases

The use of AI tools in audit work is already widespread across firms.

Standard audit tasks such as reconciliation, document review, and analysis can now be automated, said Liew Nam Soon, EY Asean regional managing partner and Singapore and Brunei country managing partner.

He added that time-consuming activities – such as sifting through lines of journal entry data – that previously took days to complete can now be accomplished by AI “in minutes”.

This frees up auditors’ time to focus on complex analysis and judgment, “where it matters most”.

AI algorithms are also “particularly effective” at identifying patterns within large datasets, flagging transactions that deviate from these patterns and detecting anomalies such as errors or fraud.

“This allows auditors to zero in on inconsistencies that demand further scrutiny,” Liew said.

Choo Eng Beng, assurance leader at PwC Singapore, said the firm has observed productivity gains through the use of AI in tasks such as document extraction and summing-ups, automating the drafting of deliverables, and running data queries on large datasets.

PwC’s FS Review Suite, for example, uses machine learning and automation to test mathematical accuracy, check prior-year comparatives, and ensure internal consistency in financial statements.

Another tool, Information Transformer, taps OpenAI’s application programming interface to extract and summarise client documents – such as board minutes and valuation reports – into formats ready for auditors’ review.

Training future accountants

As AI tools become increasingly integral to auditing, equipping the next generation of accountants with the skills to effectively utilise these technologies has become a priority.

Since 2018, Singapore Management University has been “actively” adapting its accounting curriculum to incorporate AI, said Cheng Qiang, dean of SMU’s School of Accountancy.

This includes using generative AI tools such as OpenAI’s ChatGPT as a “study assistant”, he noted.

Students leverage such tools when working with programming languages such as R, Python, and SQL. These tools help generate “boilerplate” code – pre-constructed templates that can be customised quickly.

However, Prof Cheng highlighted risks associated with adopting digital tools. A “critical” concern is bias in AI recommendations, which can arise from the data used to train AI models.

“For example, an AI analysing a loans dataset may inadvertently exhibit bias against a specific demographic if the training data reflects higher default rates within that group,” Prof Cheng explained.