Introduction

We read and hear about the need to apply Artificial Intelligence (AI)/Machine Learning (ML) in all areas to benefit from automation in all its facets. Hundreds of case studies prove its usability, and new business models arise from this technology. In the meantime, many conventional organisations, particularly their FP&A teams, feel as if they are missing the train. So why do we struggle to use AI/ML specifically in FP&A?

Technology Applications in FP&A: Pros and Cons

To better understand the complications that may prevent FP&A professionals from successfully applying AI/ML and harvesting the efficiency/effectiveness gains, we will look at two concrete applications of this technology in FP&A.

The first application is predictive forecasting, where a financial forecast is created by algorithms based on AI/ML. Statistical forecasts have been around for a while, so many of us have already been familiar with this approach. Selecting the required data and building a high-quality predictive model is often a big challenge. Explainability/interpretability of the forecast is the biggest challenge to overcome. Stakeholders interacting with the AI/ML generated forecast must know the assumptions behind the black box forecast and why this is the most likely view into the future.

What are the business reasons to believe in this unbiased machine-generated financial forecast? The argument for the model accuracy is not striking enough if business drivers and their effects cannot be explained simply and clearly. Meanwhile, the term XAI, standing for Explainable Artificial Intelligence, is gaining popularity. It has been rising from the acceptance challenge. These models are meant to provide a high-quality forecast while being able to explain the modelling and the coherences simultaneously. With significant change management efforts, company-wide acceptance and integration will likely take several years. Human and Artificial Intelligence interactions require fundamental knowledge, diverse practical experience, and general technical and business acumen.

Automation of processes and, ideally, decision automation is possible with AI/ML. For example, we can look at highly popular ChatGPT from OpenAI. It is an application that can answer every question and, if needed, provide multiple answers by learning from the feedback. Taking our title question and submitting it into the ChatGPT algorithm as one of the answers, we get the following response:

AI/ML (Artificial Intelligence/Machine Learning) can be a valuable tool for Financial Planning and Analysis (FP&A), but it is not a magic bullet. While AI/ML can provide insights and predictions that can inform decision-making and forecasting, it still requires human expertise and judgment to interpret and act on those insights.

AI/ML can automate some repetitive tasks involved in FP&A, such as data entry, cleansing, and report generation. This can free up time for analysts to focus on more complex analysis and strategic decision-making. AI/ML can also help identify patterns and trends in large datasets that might be difficult for humans to detect and make predictions based on historical data.

However, AI/ML is not a silver bullet that can solve all problems in FP&A. It requires high-quality data to work effectively, and it is important to have human oversight and validation of the results. AI/ML also has limitations in dealing with uncertainty and unexpected events and may not be able to account for factors that have not been observed in historical data.

Therefore, while AI/ML can be a valuable tool for FP&A, it should be part of a broader analytical approach that includes human expertise, judgement, and business context.

We can be fascinated with what AI/ML is delivering with a smart application and training of the model. Imagine a similar model that will answer requests from the CEO/CFO like: 

• What was our EBIT margin in the last three years?
• Visualise our sales trend for the last 12 months and predict the next quarter.
• Which cost owners are overspending in February?

It will fundamentally shift the role of profiles in FP&A. In addition, efforts invested in report / PPT generation will be significantly reduced.

At the same time, efforts in standardising the data sets and processes enabling AI/ML will increase. FP&A as a business partner with strong finance, business, and AI/ML acumen, is becoming a more popular profile.

Summary

Here we are coming back to AI/ML as the magic bullet for FP&A. AI/ML is not magic, even though some of us might think so. It is advanced algorithms and logic that we need to know and be able to explain its result. The interaction between the FP&A professionals and the AI/ML toolbox defines the success of the outcome. It is right now the time to embark on the journey, considering the learnings of the first movers and reducing your investments and efforts. In the future, we will see more advanced algorithms that will simplify the adoption of AI/ML and make the barriers easier to overcome. AI/ML will become a commodity, while the implementation of this toolbox by FP&A professionals will define if it can generate added value.