This article was originally published in the October 2024 issue of the Journal of Accountancy
To successfully assess and implement AI technology, finance needs to engage with project sponsors, adapt business case models, and identify failing investments early. Since the launch of ChatGPT in November 2022, artificial intelligence (AI) has gone from a niche technology to a purported agent of transformational change in everything from disease diagnosis health care to fraud detection. Goldman Sachs estimates that global investment in AI could reach $200 billion by 2025.
As finance professionals, we have been here before. Each new technology — such as ERP systems, data warehouses, e-commerce, and data analytics — is touted as the solution to myriad problems and the creator of new opportunities. Reality has been a little different. Each of these technologies has changed finance in its own way, but none has been the panacea that was originally promised.
The reality is that no single technology is the answer. AI may be another, potentially powerful, addition to the business toolbox, but there is no such thing as a stand-alone AI business case. There are only business cases that holistically evaluate the technology and human elements that must work together to deliver value.
As finance professionals field an increasing number of business cases relating to AI applications, here are five steps they can take to become effective partners in assessing and adopting AI across the enterprise.
Understand
AI is not a single system, tool, or technology. It encompasses any technology that exhibits some form of human-like intelligence. Examples include understanding speech, identifying patterns, predicting outcomes, solving problems, or answering questions. As such, elements of AI have been around for decades. One of the first AI applications was a checkersplaying program run on a Ferranti Mark 1 computer at the University of Manchester, England, in 1952.
AI’s current popularity has largely been driven by the exponential growth of data availability, processing speed and power, and analytic capability. These developments have unlocked numerous potential AI applications in areas such as machine learning, natural language processing, and analytic and generative AI.
As with any technology, AI is only as good as the people who build, use, and maintain it. AI can identify patterns and trends, answer questions, develop forecasts, suggest alternate courses of action, and create content, but if the outcome does not deliver value, it is of no use to the business.
Engage
Finance needs to play a delicate balancing act of carefully validating the prospective value of AI investments without being seen as a barrier to change. Finance must be the voice of reason and rationality. By engaging early with project sponsors, finance can better understand AI’s potential impact and collaborate on developing the most appropriate investment criteria and risk assessment techniques. Beyond traditional accounting roles, finance also has governance and stewardship roles to play. AI relies on the availability of accurate, timely, and trusted data, much of which falls under the finance team’s purview.
Evaluate
Simple net-present-value, payback, or internal-rate-of-return methods are useful for productivity or costreduction-based business cases, but they fail to address the innovative value of many technology investments, AI among them. AI can inform decision-making, mitigate risk, and identify opportunities for innovation. Business case models need to be adapted to reflect the variety of possible outcomes associated with AI investments. Techniques that may be useful include Monte Carlo simulations; real options analysis, which allows opportunity cost estimations of continuing or abandoning a project; and game theory, which can predict how multiple players, such as competitors, will act.
The benefits may not be directly financial in nature. For example, do chatbots and digital assistants lead to higher rates of problem resolution and customer satisfaction? Do AI-powered recruitment applications deliver candidates that are both qualified for and likely to accept job offers? Do AI-based text-generation tools deliver logical and readable text that is trusted?
Monitor
Measuring the return on AI investments requires continuous monitoring. The assumptions built into the business case may change, the business rationale may disappear, the technology may become obsolete, or the expected results may fail to materialize.
Increasingly, organizations are defining not just the criteria for success but also the criteria for abandonment for any given initiative. This requires defining the conditions under which the initiative no longer makes sense from either a strategic, operational, financial, or technological standpoint. The earlier a failing investment can be identified, the sooner resources can be redeployed to other initiatives with more potential.
Adopt
Finance and accounting are attractive areas for deploying AI, and, in many organizations, finance has been at the forefront of AI adoption. Early applications included credit scoring, customer payment analysis, and driver-based forecasting. With the increasing availability of more scalable and cost-effective solutions, the potential uses of AI are expanding quickly. In June, The Wall Street Journalreported on how Amazon is deploying AI across many finance areas, including fraud detection, contract review, financial forecasting, compliance, and tax analysis.
There are also attractive opportunities to apply AI throughout end-to-end business processes; around common business dimensions such as product, customer, and employee; and oriented to specific business decisions, such as which markets to serve and what products to offer.
By understanding, engaging, evaluating, monitoring, and adopting AI, finance professionals will be better positioned to not just evaluate business cases for AI investment but to also identify new opportunities for adoption. At one large European consumer goods company, for example, finance teamed with engineering and operations to comb through years of machine maintenance and repair data to develop predictive AI models that significantly reduced unplanned maintenance costs and optimized planned maintenance costs based upon individual machine performance. The results delivered productivity, quality, and financial benefits in less than three months.
By being an enthusiastic but thoughtful advocate for disciplined AI investment, finance teams can help their organisation’s realise the benefits of these technologies while also being mindful of the risks.
Comments