Over the last couple of years, AI has become the buzzword in business. It seems to be everywhere, usually with the mention that it will disrupt industries, sections and professions. In this article, we want to take a look at AI and auditing: has AI come to our profession as well? And what will this look like?
A recent KPMG survey revealed that four in 10 auditors expect AI-generated efficiencies to reduce the size of auditing teams. Circit’s view is that AI will come up and continue to progress and evolve, but not to replace auditors or to reduce teams. Rather, AI in auditing will mean that auditors can hand over repetitive tasks to allow for increased human attention and focus on areas where this is needed, where it adds additional value, and where it contributes to audit quality.
What Is AI?
Most laypeople will have heard of AI from the explosion and growth of ChatGPT. The generative AI-powered chatbot launched in December 2022 and reached 100 million monthly active users within two months. While ChatGPT is familiar to many, there is still a general misunderstanding of the underlying technologies AI consists of. In its broadest sense, AI is a subset of technologies that use computing power to undertake tasks that ordinarily require human intelligence.
AI in auditing will be able to fulfill repetitive tasks and undertake complex analysis that exceed human capabilities. Examples include review of financial statements, automating repetitive tasks such as data entry and full testing capabilities that go beyond traditional sampling methods. As these technologies continue to develop and become embedded within the tech stacks used by finance professionals, work will become more streamlined, resulting in higher quality and faster financial reporting.
A Look at the Different Types of AI
While we simply refer to it as ‘AI’, this term covers different models which have different specialisations.
Natural Language Programming (NLP)
NLP is one of the most common AI technologies and many individuals will have interacted with it as it powers everyday virtual assistants, such as Siri and Alexa. It can understand and manipulate human language in both spoken and written forms.
NLP is also sophisticated enough to measure sentiment, allowing it to create accurate outputs based on the emotional tone of commands. Deloitte wrote a fascinating article highlighting how Social AI can be beneficial, for example when it comes to client management and client satisfaction.
Machine Learning
Machine learning is a type of AI that enables computing power to learn from data and improve performance without being programmed. Algorithms become more accurate, making predictions in line with the volume of data they are trained on.
Large Language Models (LLMs)
LLMs can understand and generate human-like text. They are trained on vast amounts of text data to learn patterns and the context of language so they can respond to user inputs accurately and coherently. They can summarise large amounts of information and answer questions, and they are one of the main technologies ChatGPT uses.
Use Cases
Many accounting and audit vendors are already using AI directly in their products, and general AI software tools have capabilities that can support the workflows of finance professionals. We’ve gathered a summary of existing use cases for AI in auditing and accounting:
Automating everyday tasks
Leading ledger vendors, including Sage and Xero, have rolled out AI-powered copilots that can perform everyday accounting tasks. Copilots are sophisticated virtual assistants that leverage LLMs to enable users to enhance their productivity by asking context-specific queries related to existing data. Accounting copilots allow accountants to generate prompts that complete tasks such as creating invoices, chasing overdue payments, editing quotes, instant cash flows, and real-time insights. In audit, co-pilots could help streamline client communication, data entry and financial statement analysis.
This has several advantages: staff save time on repetitive tasks, saving up time to spend on elements that benefit from expert attention, such as risk analysis. Time-savings generated from faster analysis and reduction of manual tasks means auditors can provide timely advice to clients and potentially pass cost savings on as well.
Risk Assessment
AI can support audit planning processes by LLM-supported vendors using algorithms to analyse large data sets and trends to focus on high-risk areas to focus on during engagements.
KPMG are already using generative AI globally to monitor audit risk. This will enable auditors to increase the quality of audits, potentially reduce costs and free up time to question those charged with governance. Jobs can also be completed faster due to this minimising the time needed during the planning stage. Additionally, this will create the opportunity to move to a continuous auditing approach, with real-time risk monitoring evaluating live transactional data.
Fraud Detection
Traditionally, audits are completed via a sampling approach, with auditors selecting a handful of transactions above the materiality threshold. Until now, the sampling approach has prevailed due to the reliance on manual processes. However, machine learning capabilities will allow audit software companies to analyse complete data sets and flag any outlier or unusual transactions, rather than those just above the materiality threshold.
Contract Reviews
Significant audit time is expended searching through contracts, whether this is to identify clauses (i.e. covenant conditions), termination dates of suppliers and customers, or further information to determine the treatment of transactions. Looking for these pieces of information will become easier by auditors using NLP tools to identify and extract relevant information. From these findings, auditors can confirm the nature of trickier transactions and assess any associated risk.
This Is Just The Tip Of The Iceberg
While all of the above use cases demonstrate how we can benefit from AI in auditing, this is just the tip of the iceberg. AI capabilities are developing at an exponential pace. The first version of ChatPGT could remember around three pages of text, whereas a recent paper by Google claims that the company has technology to give LLMs infinite context. This creates a mind-boggling array of possibilities for analysing and processing accounting and audit data.
Individuals should apply a tinkering mindset, using the likes of GPTs, to build out tools and workflows bespoke to the needs of clients. Dudley Gould, VP of Business Development at Circit, explored some of these possibilities, including the interpretation of accounting standards, in a recent post on our blog.
Rather than reducing team sizes, AI in auditing will be an enabler of new workflows, automating and alleviating everyday tasks that take up time but don’t add value. If we view AI as a copilot or assistant, paving the way to new functionalities and capabilities we didn’t have before, we can see how auditors and accountants can work alongside this technology to add value to audit work and quality.
