Auditing Automation: Can AI and Blockchain Truly Prevent Trust Issues and Trip-ups?

Auditing Automation: Can AI and Blockchain Truly Prevent Trust Issues and Trip-ups?

Audits in general, and especially in accounting, are all dependent on trust. Can we trust that there aren’t unintentional errors or discrepancies in the processed figures? Can we trust the finance team that has put together the numbers? Can we also trust that audit committees have asked the right questions?

In most cases, the answers are “yes”. But we can be forgiven for thinking otherwise after reading news headlines like the Financial Reporting Council (FRC) issuing audit firms a record £43 million in fines last year.

High-quality audits remain the alpha and omega in inspiring confidence and trust across capital markets, facilitating economic growth and serving public interest. This is why technology implementation is being considered to elevate standards and address the following auditing pitfalls.

Accounting Automation to Reduce Chronic Human Error

Human error tends to be a leading cause of accounting errors. More than a quarter of accounting professionals previously reported that they have discovered incorrect data being manually input into an enterprise system at their firms. Accidentally deleting customised Excel formulas (cited by 17 percent of professionals) also contributed to financial reporting headaches.

‘Accounting errors’ is a term used in financial reporting to describe a non-fraudulent discrepancy in the financial documents of a company. These can include:

  • Error of omission: a financial transaction that does not appear in the documentation or is not recorded – by mistake.
  • Error of commission: a recording of a transaction for the wrong value in the correct account, such as subtracting a sum that should have been added.
  • Error of principle: a financial transaction that does not meet the international requirements and generally accepted accounting principles. It appears as an accounting mistake in which a figure is recorded in the incorrect account, thus violating the fundamental principles of accounting. It is a procedural error that consists of the correct value of the entry but placed incorrectly. These types of errors are also called input errors.
  • Transposition error: this occurs when two or more digits that are reversed (or transposed) individually or as part of a larger sequence. It appears as an error in data entry when posting a new recording. Although it is usually small and unintentional, it can result in further miscalculations, which can then lead to significant financial losses, as well as more time being invested to rectify the problem.

I’ve even seen several cases of junior accountants climbing ladders in overwhelmingly large warehouses and counting physical inventories, so that the numbers can eventually be factored into financial reporting – another practice prone to human error for pretty clear reasons.

If a company discovers an accounting error that would have an important financial repercussion, it will then need to undertake additional steps to rectify the financial recordings, and issue a statement owning to the error and releasing the correct entries. The is detrimental in the process of maintaining trust, and highly inefficient.

While the stakes in accounting and auditing are high, we must still acknowledge that human error is going to happen because we are fallible, especially when having to execute repetitive tasks for months on end. For this reason, we can’t just ask people to “be more careful” or simply blame them after an error occurs. Rather, we can turn to technology-driven tools and systems to alleviate this burden.

With the ability to streamline enterprise operations and direct the organisation’s workforce to more value-added endeavors, workflow automation or robotic process automation (RPA) has been widely implemented in tax and advisory services and is increasingly being deployed in audit.

Workflow Automation

The review of electronic documents, for example, is one important audit area that is already undergoing transformation by using RPA software to capture and interpret a transaction, manipulate data, send standardised responses and automate communication with other processes. RPA is well-suited for reconciling revenue-based transactions or other similar routine tasks, which then allows auditors more time to focus on clients, analyse data, and gain new insights from enterprise data.

With workflow automation, much of an audit’s tedium can be reduced. Rather than burning the midnight oil lost in stacks of paper, audit professionals can experience higher-level activities sooner in their careers.

AI-Powered Auditing to Negate Sampling Errors

Auditors use sampling mainly because they are not required to seek ‘absolute certainty’ – instead, they are looking for ‘reasonable assurance’. They therefore always try their best to pick samples or batches of transactions that will help them attain reasonable assurance that the rest are legitimate.

However, it is not always possible to get a 100 percent representative sample, possibly due to bad luck or pure errors in judgment. To the layman, the lack of ‘absolute certainty’ doesn’t bode well for trust.

Many believe that auditing every single transaction is ideal, but that aspiration has been extremely resource-intensive and impractical, until the advent of Artificial intelligence (AI).

AI involves the development of algorithms and software that’s able to perform tasks that normally require human intelligence. And because AI can tackle many time-consuming and repetitive tasks traditionally performed by humans, it can enable an audit to avoid the typical trade-offs between speed and quality.

Two AI technologies that are especially relevant to audit are natural language processing (NLP), which enables a system to read and understand key concepts in electronic documents, and machine learning, which enables a system to ‘improve itself’ without being manually reprogrammed.

With enterprise-grade incarnations of Siri and Alexa – such as SAP Conversational AI – making its way into professional accounting, an auditor of the future could very well ask a voice assistant in his workspace to “look up the invoices from 2019 between companies A and B and find this specific item.” The auditor might even be able to review every document or invoice by engaging numerous software ‘bots’ with one voice command like “please audit all the invoices and come back to me by tomorrow morning.”

Dataset Analysis AI

As audit evidence increasingly becomes more digitised and stored on cloud-based enterprise resource planning (ERP) systems, computer systems at accounting firms can now interface with an audit client’s systems to transfer and compile data automatically. The injection of workflow automation and AI could then supercharge the analysis of datasets, test 100 percent of a company’s transactions, and bring us much closer to ‘absolute certainty’.

Blockchain to Mitigate Inherent Conflicts of Interest

Finally, we are often taken aback by failures because we incorrectly assume that audits of company financials have been conducted by independent audit firms with no financial ties to the companies that they are auditing and that auditors have no financial incentive to maintain that relationship.

In reality, auditors are inherently under pressure to stay in their clients’ good graces, since the clients are the ones paying the bills for their services and essentially putting food on the table. On the other hand, clients are inherently pressured to ensure that they are putting out financial statements that meet the expectations of investors and other stakeholders.

Consider a situation where an auditor spends time verifying whether stored data has been edited through fraud or error, only to discover discrepancies but be forced to awkwardly toe the line when the client disagrees regarding guidance.

It’s not to say that auditors lack integrity. Rather, the issue here is systemic. And given that this system is endorsed not just by audit firms themselves but the entire accounting ecosystem, we must consider how we can finetune the process to enforce shared responsibility, maintain good corporate governance, ensure objectivity and reinforce trust.

So, here’s where blockchain can come in.

A blockchain is a digitally distributed ledger of economic transactions that contains two types of records – transactions and blocks. Blocks hold batches of transactions, and since blocks are time-stamped and linked to previous blocks, they cannot be tampered with without the knowledge of everyone else with access to the ledger.

Blockchain Auditing

Recording transactions immutably on a blockchain would create an audit trail for regulators to verify auditing accuracy and compliance. With blockchain, regulators will be given access to read-only, near real-time updates to private blockchains of organisations’ financial records. This would allow regulatory bodies to then play a more proactive role and analyse information if need be.  In other words, blockchain has the potential to ‘systemise’ more inherent checks and balances, thereby improving the quality, accuracy and trust in the audit process and outcomes.

Technology Still Not the Silver Bullet

While emerging technologies will no doubt enhance trust and efficiencies in audit and corporate governance, they also bring with them new challenges. AI algorithms, for example, will need to be sufficiently tried and tested before implementation to avoid inherent biases. Blockchain, as another example, would still not address the issue of transactions being omitted from the ledger in the first place.

At the end of the day, an acknowledgment of our shortcomings, as well as an understanding of the benefits and risks presented by emerging technologies, is required for auditors, management, and audit committees to most effectively leverage new capabilities and discharge their respective responsibilities. That said, this will bring us much closer to absolute certainty but never make us ‘absolutely certain’.