top of page
  • Writer's pictureInchmead

Impact of ChatGPT on finance teams – hype or reality?

As the world continues to find a place for ChatGPT, its immediate impact on the finance function will not be as great as some are claiming, argues Eugene Vyborov, CTO at YayPay

ChatGPT and generative artificial intelligence (AI) have rightly created a lot of buzz over the past few months, opening the eyes of consumers and businesses alike to the significant benefits of these technologies. From developing code to writing jokes, AI innovations will change the world as we know it, and we are already starting to see this happen.

However, ChatGPT is only one form of AI and other types have already been widely adopted by many organisations. The finance function is one business unit that has been ahead of the curve in adopting traditional AI, which consists of simple rules programmed to perform specific tasks, and automation to enable huge optimisation across basic processes.

As the world continues to find a place for ChatGPT, its immediate impact on the finance function will not be as great as some are saying. Instead of getting caught up in the hype, finance leaders should continue to invest in machine learning (ML) and automation to further increase efficiencies and protect cash flow.

Leading the AI pack

Traditionally, the finance function has been a heavily manual and repetitive job. It requires a lot of time and effort from team members to input data and chase invoices. In fact, our research found that 58% of finance teams spend more than 10 hours a week processing invoices and administering supplier payments. This is a significant amount of time spent on processes that could easily be automated so that teams can focus on more complex tasks.

The good news is that the adoption of AI and transition to full automation has been a priority since 2019 and 66% of finance teams expect to be fully automated within the next three years. As teams increasingly automate accounts receivable processes, such as invoices, human error will also be dramatically reduced.

Continued investment in AI is not only essential to reducing mundane and manual work, it can transform a business. A survey of accountants by Intuit QuickBooks found that firms who do not adopt AI risk stunting business growth (43%) or losing clients to competitors that offer richer insights (36%).

We are already seeing finance teams put AI into practice, with the technology increasingly used to create predictive customer insights. ML-powered algorithms can access historical payment behaviours to create real-time and accurate insights into how businesses can expect their customers to pay.

This deeper understanding of customers can be utilised by companies to predict when invoices will be paid and so protect their cash flow from fluctuations.

The language of machines

One key reason why ChatGPT has taken consumers and businesses by storm is that it is a Large Language Model (LLM). LLMs are trained deep-learning models that process and generate text that looks more natural and less prescriptive.

This form of AI goes beyond just automating basic data entry or the delivery of invoices. It can gather huge amounts of complex data and summarise, generate and predict new content. This could be game-changing for business units, such as the finance function, that handle large data sets, as it can unlock previously unknown insights and forecasts.

AI is already being adopted by finance teams for its data crunching ability, but an LLM can formulate this information into an interesting and digestible format.

For example, teams will be able to use LLMs to extract content from PDF pages and analyse contextual information about customer accounts. Traditionally, an employee would have to open PDFs individually and spend hours scouring them for the relevant information.

Now, AI can do this instantly and make connections between payment patterns that could easily go unnoticed by humans. While this analysis will give teams more in-depth information for forecasting, humans will still be needed to check what the LLM is producing. The LLM is dependent on what data you share with it and train it with, which means it cannot be treated as a perfect source of truth.

Ultimately, LLMs will enable a lot of additional automation that was not previously possible. Many of these use cases will involve human collaboration, where the AI helps the human to do more with less.

LLMs will impact the whole finance chain, to the point where an AI agent will be smart enough to replace most credit functions. But it is important to remember that humans will never be redundant. We will always be needed to provide checks and balances and approve what the LLM produces.

One of the most beneficial functions we will see from LLMs is in operation-style communications such as dispute resolution. For example, if a buyer always disputes invoices because they are not in a particular format, the AI will be able to automatically change the invoice into the right format. The ability to understand this issue, correct it, and then communicate this change will be transformative.

However, we are still a long way off LLMs being able to communicate in this way.

As the many ChatGPT ‘fails’ have shown us, LLMs can predict what word should come next based on context, but they cannot communicate with the same nuance as humans. The finance function often acts as the first line of customer service, so simply cannot risk AI not quite understanding a customer’s tone of voice.

Future of the finance team

The prospect of advanced AI and LLMs is exciting for many businesses, and as AI continues to advance at such a rapid pace, the possibilities are endless. Eventually, AI will be able to do anything we ask of it, but we are nowhere near that reality yet.

There is no guarantee that the text LLMs produce is accurate or relevant, and so businesses must not treat AI as a silver bullet for customer experience. And so, in the short term, it is unlikely that ChatGPT and LLMs will have a significant impact on the day-to-day operations of the finance function.

Unfortunately, this is compounded by the main barrier to AI adoption – even traditional models - still being human habit. The more experienced we are, the slower we are to change these habits.

But for finance teams to take advantage of the technology we have available today, let alone that of the future, there needs to be a mindset shift towards adoption. Those who are slow to adopt the AI and ML tools we have today, will be the ones left in the dust in the future.

bottom of page