![Product Manager Eva Michaelsen sitting infront of a computer.](https://eye-share.no/wp-content/uploads/2023/09/AD0A1505-730x450.jpeg)
Getting knowledge about the hows and whys of AI may seem like a daunting task. Even so, this is one of the most important tasks you can take upon yourself. The digitalization of accounting processes, and especially the P2P cycle (purchase-to-pay cycle) and accounts payable process, are becoming ever more important.
As a CFO it is crucial to have basic knowledge about AI to navigate the landscape comfortably. Let’s have a look at what AI is, the subsets of AI, and why XAI (explainable AI) is the best option for different kinds of ERP and accounts payable systems.
What is AI? A Short Overview
In short, AI is about creating programs and training computers to do tasks that demand human intelligence. The idea of creating machines with human intelligence is said to have originated with Alan Turing. The field made its breakthrough at a workshop at Dartmouth College in 1956, which is often marked as the beginning of modern AI.
The holy grail of AI is General Artificial Intelligence, but we have yet a long way to go to reach that goal. In the meantime, most of the AI being used in software today, including the LLMs (large language models) like GPT which many people now identify as AI, belong to a subset of AI generally called NAI (Narrow AI). NAI is made to solve not all tasks, but very specific, or “narrow”, tasks that involve problem-solving, and decision-making. The learning done by an NAI model is specific. It cannot be transferred or generalised to other models, as there is no “understanding” involved – NAI is all about computation and statistics.
Why AI in accounting?
Digitalization of the P2P cycle promises great rewards in the form of more time for employees to focus on core business and greater accuracy due to less error-prone processes. The automation of tasks leads to scalability and lower operational costs, and as your company grows, the help of AI becomes more and more important to process large amounts of data in a short time, draw reports, and keep track of the cycles of the P2P process.
It is possible to automate a lot of the process without AI, but adding AI to the mix gives even better results. With the help of AI, you can handle huge amounts of data in seconds and say goodbye to manual handling of information, you can structure data that earlier has been unavailable and unstructured, making informed decisions part of your everyday job. Tasks that used to be meticulous and tedious can be automated, setting rules not only in the past but for future reference.
There are two different types of narrow artificial intelligence you should be aware of as a CFO. These are black-box and white-box, or explainable AI.
Why should you avoid Black Box AI for Accounting?
Black Box AI is a term used to describe a kind of algorithm that sets its own rules, having learned from the training data it has been given. There is a great deal of debate around these kinds of AI systems, as the decision processes of the system are not transparent; decisions are made deep within the algorithms, and what has been done is not necessarily visible or even traceable. This means that sometimes the result may seem right, but the decision has been made under the wrong circumstances, and the user has no insight or control or even possibility to influence the decision made. It is, in other words, impossible to understand what variables have been used to make a decision.
Why is XAI (explainable AI) the Better Choice for Accounting Systems?
XAI is short for explainable AI. In comparison to the black box AI mentioned in the previous chapter, with explainable AI it is always possible to track the rules that guide the choices made in the system. With such an AI system, there will always be certainty about whether or not a decision is based on incorrect circumstances or data. The decisions are traceable and the model will make it clear how decisions and predictions have been made. In short, XAI leaves you in full control and makes the output from machine learning algorithms trustworthy, because it is easier to understand and detect errors.
In accounting, and especially in the P2P cycle, traceability and explainability are important, as they ensure that you don´t lose control over the processes running in the background of your systems. All suggestions and automated actions that the system makes will be explainable, and when corrected, will help guide the further actions of the XAI model.
AI is the Future of Accounting
AI is being embedded into the very fabric of most accounting systems, whether it is an ERP system or an invoice processing system, and as a CFO you need to make sure that you can make informed decisions regarding the use of AI in your business. Whether you go for a system with a Black Box a White Box, or an XAI version of AI, will have a great impact on how reliable and trustworthy the system potentially is.
Would you like to learn more about how AI can optimize the P2P process in your business?
We have created an ebook that covers everything you need to know about Procure-to-Pay, from basic concepts to choosing the right system. Download the ebook “The CFO’s and Accounting Manager’s Ultimate Guide to Procure-to-Pay (P2P)” for efficient tips and tricks.