Part 4 in the “Understanding IFP&A” series
Artificial intelligence (AI) is one of those terms surrounded in mystery. It’s also a term that is easy to misunderstand relative to its promises, particularly in reference to FP&A.
We might assume that AI will replace slow, biased, human decision-making and deliver instant, fact-based, automated rational decisions whose outcomes are supposedly more accurate—and hence more profitable—than the human equivalent. The only problem is that the “right decisions” often depend on an unpredictable future for which the appropriate data is unavailable.
Google defines AI as “the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.”
Wikipedia defines it as “… the study of any device that perceives its environment and takes actions that maximize its chance of successfully achieving its goals. Colloquially, the term ‘artificial intelligence’ is applied when a machine mimics ‘cognitive’ functions that humans associate with other human minds, such as ‘learning’ and ‘problem solving.'”
For FP&A, AI is the automation of actions based on analysis of available data in order to mimic the way a person would make a decision. And there’s the problem.
Yes, humans make decisions based on data, but they are also heavily influenced by experience, gut feeling, biased views, whimsical notions, and sometimes “just to be different”—none of which can be codified for a machine to perform.
Similarly, AI systems can make decisions only from information presented, which could be inaccurate, incomplete, and out of context, thereby leading to the wrong conclusions. Even then, AI will rely on some form of trend and/or correlation guided by human supposition. Identified trends will end at some point, which the data can’t predict. Nor can the data predict when the next business disruption will appear that will change the rules.
But having said all this, I believe that AI does have very real business value. Here’s are some examples:
Testing hypotheses: A hypothesis is a supposition that certain events or variables are related. AI technology can confirm or even discover them, which can then be used to predict future outcomes. The value comes when managers challenge whether these correlations and predictions “feel” right and to promote discussions about what would prevent those results from occurring.
Speeding up decision-making. Human decision-making requires the mind to assimilate the data and, based on past experiences, make an informed decision about the actions required to realize the predicted future. These actions typically come from a “rule” establishing what the organization can do. Technology can go through this process far better than a person because of its speed and ability to process vast amounts of information. It can also do this on a continuous and consistent basis. That doesn’t mean to say that AI will be right, just as people are not always right.
Triggering data-led actions in real time. Consequently, AI can be used to trigger actions, whether real-time detection of online fraud or replanning parts of the organization based on adverse results. This can save a considerable amount of time through which the system can “learn” about what works and what doesn’t.
In short, the whole purpose of AI is not to replace humans, but to be used as another tool that can help management teams clarify their thinking and make better-informed decisions.