Ready for a new frontier in the digital finance function?
“What’s happening with artificial intelligence right now is an inflection point for humanity, and a tipping point for human evolution,” according to Nina Schick, author and all-round global authority on generative AI.
From the boardroom to the dinner table, we’d wager the debate about artificial intelligence has dominated a conversation or two over the last twelve months.
But what’s behind the watershed moment for generative AI in particular?
Recent mass adoption of generative AI can be attributed to tremendous advances in deep learning. Coupled with lightning-speed developments in foundation models, a world of opportunity has opened up, says Schick:
“Foundation models are extraordinary because they’re being trained on vast datasets - we’re talking almost the entirety of the internet!"
"It makes them general purpose models capable of a multitude of tasks. They’re exceptional because the way we interface with them is speaking to them in human language.”
Where once programming language was seen as the domain of tech-geniuses and whizz-kids in IT, today, natural language processing is the new programming language, and we can all stand to benefit as a result. In other words, talking to the machines the way you might bounce ideas off a colleague, or instruct a team member to complete a task, has quickly become the new normal.
It’s no surprise, then, that when it comes to the foundational model with the most impact on the future of finance, LLMs (aka. large language models) are second to none.
LLMs interpret, summarise, generate and query vast amounts of text, including human language and data.
If ChatGPT is springing to mind, you’re on the right track. In fact, Schick predicts our collective introduction to ChatGPT will go down as a pivotal moment not just in the history of artificial intelligence, but in the history of technology as a whole:
“This was the moment advanced AI capabilities were showcased in an accessible manner to the rest of the world. And how did the world respond? With a hundred million users of ChatGPT in two months.” To put that into context, it took Netflix two and half years to reach a million users, Facebook ten months, and ChatGPT five days to reach double that amount.
Whether we know it or not, we’re living through major milestones in digital finance transformation as they happen. The difference this time around? A shift from what Schick terms an information revolution, to an intelligence revolution instead:
“Emerging technologies of the past thirty years such as the internet, the smartphone, and the cloud have all been deeply transformational from a business level, to a geopolitical level, all the way down to a personal level. I call that the information revolution. Now, these emerging AI capabilities mark the beginning of the intelligence revolution, where machines are adding exponential value to human knowledge as a whole.”
If you’re wondering where the future of finance fits into all this, you’re not alone.
According to Schick, conventional wisdom was that deep knowledge industries like finance would be impacted last because, as anyone who has tried the likes of ChatGPT will tell you, LLMs are hardly without their accuracy problems:
“However, finance is a great example of an industry with large volumes of unstructured data which happen to be in natural human language. Therefore, even when there are inaccuracies, we’re seeing that the value-add is so great as to outweigh the impact of this,” says Schick.
That value-add lies in the versatility and autonomy of LLMs, and what Schick terms “unexpected innovation” as a result of being able to sift through otherwise incomprehensibly large volumes of data to uncover fresh insights.
Still sceptical? Schick is used to people questioning whether AI capabilities could be over-hyped. After all, artificial intelligence is still just that: artificial, not sentient.
“No, I don't think AI is sentient (if you can pull the plug out and it stops working, it's probably not sentient at all). But to me it’s a moot point anyway."
“No, I don't think AI is sentient (if you can pull the plug out and it stops working, it's probably not sentient at all). But to me it’s a moot point anyway," says Schick.
She cites ChatGPT-4’s upper 90th percentile result for passing the bar, alongside a similarly impressive performance when put through the CFA.
In other words, AI might not be sentient, but should we care all that much that it isn’t?
For every doubter, there’s surely a doomsayer or two to compensate who, far from underestimating AI capabilities, might be inclined to overestimate them instead: namely, those concerned that AI could wipe out the workforce and, ultimately, the human race.
“The AI death narrative is very compelling and it’s one the media loves because it’s an idea straight out of science fiction.”
“But while it’s important to consider the existential risks, it's equally important to distinguish between near term risks and hypothetical risks.”
For now, rest assured that rolling out an LLM to better service your digital finance function is a far cry from granting AI the ability to kill us. According to Schick, most AI researchers don’t believe existential risk to be a threat worth worrying about.
Instead, Schick warns that fuelling such narratives could be taking our eyes off the ball of more pertinent and prevalent challenges. Alongside valid concerns around data accuracy in digital finance transformation, cybersecurity, and IP infringement, Schick argues a more pressing issue is the need to rethink what skills will be needed for the workforce:
“I have no doubt the way work is done is going to be severely disrupted thanks to AI. But rather than replacing jobs, the disruption will be a demand for people willing to work alongside these systems, and for the people who can see beneath the hood of AI technology. We must prioritise education and training - it’s a critical area that’s been neglected so far.”
Nowhere could this prove to be more important than for CFOs themselves:
“Imagine you have a bespoke LLM model trained on the historical data of your business, with deep domain knowledge and access to years of structured and unstructured data from your organisation. Now imagine being able to access that at the point of decision-making. Rather than threatening the role of senior leaders, that’s going to be the most valuable tool a leader can have.”
Above all, Schick’s advice for senior management is to be bold and take the reigns of AI transformation in finance into your own hands:
“AI for the digital finance function comes with huge opportunities and risks. It’s our burden, as well as our responsibility and our privilege, to shape and guide these systems."
"Yes it can be scary, but it’s also really cool! That’s how most of life is. It’s up to us to decide how the AI revolution plays out.”
Chris is the founder and MD of GenerationCFO.com and creator of the Digital Finance Function Model and a contributor to many articles on our platform. Chris focuses on the shift toward digital transformation in accounting and finance, shows you what good looks like, then helps to get you there!