Torben Anderson, Director of Engineering at Pleo, dives into the realm of Generative AI and untangles its common misconceptions, presenting its positive role in the finance industry.
We saw that 2023 was a big year for Artificial Intelligence, as everyday people were suddenly able to get hands-on with the technology. It’s fair to say that they were excited by what they saw, with ChatGPT, for example, racking up over 180 million monthly users and 100 million weekly active users, helping people do everything from writing wedding speeches, appealing insurance claims, planning workouts and more.
Despite this, we still don’t know what AI’s full capabilities are. But instead of treating this blind spot as something to worry about, we should approach it with curiosity. Because while there are things we should be aware of, these are nothing compared to the value we’ll ultimately get from it. Something that is especially true when it comes to finance.
Finance AI is worth our time
According to the CFO’s Playbook for 2024, business decision makers aren’t convinced about a role for Generative AI in finance. Just 27% agree that they are confident about the introduction of AI into their finance operations, while less than a third (29%) agree that AI could play a role in reducing the time spent on admin off CFOs and finance teams and returning it to more high-level strategic tasks.
What’s confusing though is that the finance sector is no stranger to the value of AI – helping automate payments, deposits and transfers, enhance security and communicate with customers via chatbots. These activities have shown AI to be capable of streamlining the performance of finance teams, so why aren’t we jumping at the chance to implement its advanced cousin? One reason could be the confidential nature of the sector. But finance leaders should know that this isn’t about handing customer details over to the machine. It’s about identifying where AI can outperform and assist humans and find the right home for it in your business. With this in mind, here are some steps to get started.
Don’t be too concerned
Bringing AI into your finance operations shouldn’t be something you sweat over for weeks. It’s helpful to be cautious, but not to the detriment of your business. A good rule is to treat it as though you were hiring a new person.
One of the key worries about Generative AI is the technology’s tendency to ‘hallucinate’, e.g. when it’s asked a question and returns an incorrect answer with perfect confidence. But this is really the crux of what makes the technology work. All it does is give you what it considers the most likely answer, with no way to tell you how confident it is in that answer as of yet. Not unlike if I give a colleague incorrect information, despite being sure it’s correct. However, there is a lot of research in this area, so we can expect it to become less of a worry in the future, plus there are already a lot of good techniques to mitigate risk. Ultimately, these traits are why you’ll be hard pressed to find a company that isn’t pairing AI with human input. For instance, letting the technology do the writing and your human talent the proofreading.
Do a slow roll-out
The AI space is shifting every day – at a faster rate than any other tech sectors – and the tool you favour for solving your business problems today might change tomorrow. This is why you shouldn’t hitch all your horses to one vendor. There are some companies doing fantastic things with AI right now. But for many, the truth is that you could get most of the value you’re getting from the Generative AI-based tool you use straight from a solution like ChatGPT instead – if you know what you are doing. Most of the tools and automation are simply layers on top of this. So, building these so-called ‘prompting’ skills is the key you’re looking for.
To do this, businesses should avoid costly subscriptions to multiple services and instead provide everyone in the company with access to AI like ChatGPT or Google Bard as fast as possible. There are plenty of safe AI playgrounds for your teams to practise in and get up to speed with, so be sure to use them.
Embrace enthusiasm
The best thing you can do to help AI grow organically within your business is to find the people who are naturally drawn to it. You need as many people as possible who want to work and experiment with it and to find those ‘aha’ moments that can change how we work. It’s crucial that they feel empowered to experiment though, which requires a green light from management, preferably all the way from the C-suite.
When it comes to those who don’t want to use it, you shouldn’t focus your efforts on convincing them – not yet anyway. This can happen organically through seeing in-house use cases. For instance, McKinsey states that Generative AI will unleash the next wave of productivity for workers. So, in the office, if someone’s neighbour is delivering a better performance and maybe getting a bonus as a result, this will go a long way to convincing AI naysayers to get onboard and reappraise the initial perception they had.
Find your own route to AI implementation
Whatever your route to AI implementation, one constant should be enablement. Ultimately it should not be seen as a tool to do everything. Find specific use cases where it’s a great fit, prove the success and then move on to the next thing. One example of a good place to start is where someone has to read a lot of text to find and present small parts of the same information. AI excels at this and can do it in a fraction of the time.
At Pleo, we’re driving ahead with Generative AI with our own in-house version of ChatGPT. We’re a company that is investing in AI, but one that wants to do it properly. Finance teams aren’t incompatible with AI, but the subject clearly raises concerns. So long as we don’t dismiss them and instead overcome them, then we’ll remain headed in the right direction.