“Don’t be like the person pushing a boulder uphill,” KC Rottok Chesaina, Chief IFRS Officer, Financial Minds, cautioned the finance industry at the recent Finance Indaba in Sandton. “Be curious about AI and take time to experiment with the technology in financial reporting.”
KC presented his views on the applications of AI in financial reporting during a lively, well-attended session. Urging the audience not to ignore the various technologies at their disposal, from ChatGPT through to Claude, Complexity, Deep Seek, Notebook LM and Microsoft’s Copilot, he said the business world needed to discard the idea that asking AI for support or information makes anyone less of a professional.
AI tools are giving finance professionals a competitive edge at work and can assist them in the various stages of financial reporting: summarising complex agreements, calculating financial metrics, generating journal entries, drafting accounting policies and conducting preliminary research.
KC unpacked the application of AI in financial analysis by presenting a case study on a power company’s transformer lease, tackling the question of whether it constitutes an asset under IFRS 16 or not.
“By uploading documents to enterprise-grade AI tools like Copilot, financial professionals can swiftly extract key transaction elements, analyse lease classifications, generate preliminary calculations and recommend how to record an item or transaction in the accounts,” he explained.
“But the onus remains on the finance professional to interpret the results and make the financial decision,” he said. “AI only presents the arguments for and against classifying the transformer as an asset.” Once the professional has taken a decision on the correct manner of recording the lease, they can use Copilot to calculate the Net Present Value (NPV) thereof and use it for their amortisation schedule.
Risks and ethics of AI in reporting
Touching on the risks involved in using AI in financial reporting, KC reminded users that AI tools can hallucinate, generating plausible but factually incorrect data, figures or IRFS references. For this reason, users must always verify AI-generated outputs against their own knowledge and other authoritative sources.
There are a number of ethical considerations regarding AI tools in financial reporting. Sensitive data can be exposed when inputting financial statements or work-papers into AI tools, and over-reliance on AI may lead to breaching the Saica code of ethics, which requires objectivity and professional competence.
KC warned that “ChatGPT does not interpret IFRS paragraphs in the context of specific transactions, and AI responses are based on training data”. He advised that enterprise-grade AI tools like Copilot are more secure solutions for the finance sector compared to public AI platforms.
Emphasising that AI can remove the mundane tasks from the finance professional’s workload, KC enthused that machine learning is a fantastic tool to deploy for repetitive tasks – any function that doesn’t require deep thinking.
