The effective use of advanced analytics and AI is hampered by data quality problems, analytical capabilities and difficulty in integrating multiple data sources. Such was the observation of a report published by the ACCA and Chartered Accountants Australia and New Zealand in July, based on a global survey of 1,600 finance professionals.
If companies are going to implement AI into mission-critical processes, finance teams must be able to trust the data underneath it.
Most finance organisations have data coming from a lot of different databases and files: the ERP, financial close systems, financial planning and analysis platforms, treasury systems and other applications across the business.
The challenge is not necessarily that the data does not exist – the risk and concern is whether that data is consistent, connected and retains the right metadata as it moves across automation connectors and related systems.
Unreliable data gives unreliable answers
“Companies can implement the best AI in the world on top of unreliable data and the financial results will yield an unreliable answer,” says Omar Choucair, CFO of financial software provider Trintech. “That is why integration, governance and controls must be part of the foundation.”
His view is that if finance teams are not confident that the data is reliable, that they have the right controls around it and can explain the output and that it could survive the rigors of an audit, they are not ready to rely on AI for a critical finance process.
According to the report authors, critical thinking and sceptical validation are essential to reduce automation bias, anchoring bias and deskilling risks.
Choucair agrees that finance teams should seek AI use cases and implement financial controls around what it can do. He adds that this has to be grounded with discipline around data and accuracy validation.
Choucair also observes that organisations will naturally have different levels of expertise and comfort with AI – and that while some employees will adopt new tools quickly, others will need more training, examples and support.
“Simply giving people access to AI does not create transformation,” says Choucair. “Companies must help employees understand where the technology can genuinely improve their work and where human judgement still matters.”
Collaboration between finance and technology
Another consideration is tighter collaboration between finance and technology. Finance teams do not need to become technologists. However, they do need sufficient understanding of AI to challenge assumptions, evaluate risk and determine where it can create value.
Likewise, CIOs and technology teams are spending more time understanding controls, governance and accountability required in finance, says Choucair, adding that AI investments are no longer purely technology decisions.
“The technology team needs to understand what is possible, while the CFO needs to understand the business case, the upfront and maintenance costs, the controls, the workforce implications and ultimately the return on the investment,” he says.
Choucair believes digital transformation and AI investment in the office of the CFO will continue to evolve quickly. “Organisations that get the most value from it will be the ones that combine that innovation with trusted data, strong controls and upskilled employees who understand the new AI applications and how to use it responsibly,” he concludes.
