If I asked you how old ChatGPT is what would you say? Despite feeling like a relatively new software, ChatGPT launched in June 2018. But it wasn’t until the launch of GPT-3, where people could directly interact with the software, that it quickly became the fastest-growing consumer software application in history.
From being a piece of software to tell a funny joke, ChatGPT now has wide-reaching influence on customer services, education, content creation and supporting businesses by automating tasks.
The history of ChatGPT highlights to us how rapidly technology can change our lives and quickly become something we can’t live without. Of course, ChatGPT is not the only rapidly growing AI software out there. Competitors like Claude, Gemini and Llama are out there and developing as quickly as ChatGPT. With this competition comes plenty of advantages, but it also brings its challenges.
So today we will focus on the use of AI in business and its associated risks and benefits. Note, these risks and benefits are not limited to ChatGPT but can be considered for many types of artificial intelligence out there that is designed to support businesses.
Using AI systems that support with manual data entry, such as invoice processing software, can help to speed up processing of large volumes of data. AI systems like this have an added benefit of automating repetitive work such as bank reconciliations and invoice matching, giving accountants and business owners more time to focus on analysis of the numbers and related advice.
The faster processing times also help to speed up month-end processes, helping to produce more meaningful and timely results for business owners.
Using AI to handle large data entry and repetitive tasks gives businesses financial results sooner than if they were carried out manually. This can help with more timely decision making and making positive changes to the businesses sooner than would have previously been possible. For example, if a business’s bank account is connected to an AI accounting platform, then it can see in real time how much cash the business has, who it owes money to, and predicted cash flows for the weeks ahead.
AI systems can now handle vast amounts of data and process that data accurately and efficiently. Often, if a human was to process a similar amount of data, then not only would it take a lot longer to process, but it would also be at risk of errors from manual input mistakes, errors made in calculations, and other errors such as transposition errors.
To help further the accuracy of that data, the time saved by not processing it manually can be used to analyse that data and spot any potential errors made by the automation software.
AI is now capable of alerting businesses to real-time fraud detection. For example, an employee expenses software can set up a parameter on the amount an employee can spend on subsistence. If an employee submits a claim above these parameters, the software will alert the user to this where the user can determine if it is a business expense.
This type of software is an example of how AI can aid businesses with fraud detection across employee expenses, invoice processing, payroll and bank transactions.
Despite its initial set-up and implementation costs, the use of AI can be a cost-saving measure. Automating tasks helps to reduce labour hours and improve efficiency for businesses. This can have quite a significant cost saving particularly for smaller businesses.
With cost savings comes the risk of job displacement. Now that AI is more than capable of handling vast examples of data-driven tasks, this could reduce the need and availability of more junior level work for accountants.
AI is pushing the need to revisit the accountancy profession to ensure new accountants are equipped with the latest skills, which is more often moved away from data-driven tasks and towards more value-adding activities provided to a business.
As touched on earlier, the implementation of AI systems can be a costly and time-consuming exercise. The costs and time required in teaching the software, integrating it with a business’s systems and training users can quickly add up and act as a barrier to entry, particularly for smaller businesses.
The quality of data outputs is only as good as its inputs. If businesses or accountants are inputting wrong, incomplete or poor data then the results produced by the AI will also be of low quality.
Using AI software for the first time can also have its challenges. AI outputs are often based on patterns in the data, so if a business is using AI for the first time these earlier ‘patterns’ can be somewhat bias or less reliable. For example, if an AI tool has been used for partner expenses, it might typically put a meal expense to client entertaining, but if that software is then used for employee expenses, it might code meals out to client entertaining when in fact it is more likely to be subsistence spending.
These types of risks can be mitigated by continued used of the AI software where over time the accuracy and reliability of that data can improve.
Arguably one of the more serious risks of using AI is the potential for data breaches. By their intended uses, AI software processes and stores sensitive client data that, if a breach was to occur, could expose confidential information. Such data could include, names of customers and suppliers, company account balances including bank balances, and names and details of employees of the business.
However, there are ways to help limit the potential for hacking and other breaches such as by using Two Factor Authentication to access software, only accessing it on company devices and under protected WiFi, amongst others.
All these risks highlight the need for regulations. However, as we’ve discussed with ChatGPT, the developments in technology far outpace the introduction of regulatory and compliance changes.
Regulators require transparent and traceable data, but AI models can’t always provide this. For example, if an AI is relying on complex algorithms to determine how it categorises a business expense, then how can a business explain this to a regulator? This shows the difficulties in compliance with regulators when using AI as there is likely to always be this gap between the requirements of regulators and what the AI software can provide.
We hope this blog has shown to you the many advantages of adopting AI in your business and shone a light on the potential drawbacks of AI and what you need to be aware of to protect your business. If you have any questions on what has been discussed in this blog, feel free to reach out to us.
