If there’s one topic we all can’t stop talking about, it’s artificial intelligence. Across platforms, industries, age groups, and dinner tables, AI’s enormous potential — and pitfalls — is dominating conversations.
What does the development of AI hold for finance and banking professionals? How is AI being used today and what can we expect in the future? Can we maximize its power while minimizing its risks? And how exactly can we work with AI to boost our performance and productivity?
In our recent panel discussion, Bots, Bard, and ChatGPT: How Finance & Banking Professionals Can Capitalize on the Latest Advancements in AI,CFI’s VP of Content, Ryan Spendelow, and Director of Business Intelligence, Joseph Yeates, got right into it. Read on for some key takeaways and sign up below to watch the full 60-minute session for free.
What Is Artificial Intelligence?
Artificial intelligence systems are designed to mimic human intelligence and learning. There are two main categories of AI: general, which can theoretically be applied to many domains to solve any problem, and narrow, applied to a specific domain or problem. Within that, there are many different AI components.
Narrow AI is far more common and typically what you’re thinking of when it comes to AI tools. In finance and banking, this could take the form of a “robo-advisor” introducing customers to different banking products or automated fraud detection.
“Natural language processing is a really interesting one because it’s a way for AI to interact more naturally with humans,” says Ryan Spendelow. “It combines linguistics, machine learning, and math to really understand text and spoken words — things like ChatGPT, Google Translate, Siri, and Alexa.”
Machine learning is an important aspect of AI technology, contributing to many components like neural networks. Neural networks are particularly relevant at the moment, Ryan explains, because of their path toward deep learning.
“A neural network is a bunch of algorithms designed to simulate the human brain… with nodes firing messages to each other. Deep learning is where you have an input and output layer of nodes, with hidden layers in between to increase the complexity of operations.”
ChatGPT is an example of how interconnected AI technology and components are. As a natural language processing model, it uses neural networks and deep learning to provide a response to the words you type in. A few members of the CFI team use Finchat.io, which is like ChatGPT for financial analysis. “I can tell it to give me a margin analysis for Microsoft over the last 5 years, and then have a follow-up breaking it down by quarter,” says Ryan.
There are endless use cases for AI across every industry, but let’s dig further into how finance and banking professionals are leveraging it.
Joseph kicks off with a framework for thinking about AI and how these different components and products work together — comparing how important a task is vs. how difficult it is. For instance, filtering spam is lower on importance and difficulty, and based on simple rule sets that still learn from human feedback (telling our inbox to flag or unflag something as spam).
“As we start to talk about AI in finance, the two key takeaways using this framework is that we want to automate and regulate the lower importance work, which will allow people to focus on higher value work,” Joseph explains. “So we can leverage what AI is good at and focus on what humans are good at that maybe AI isn’t as strong in at the moment.”
While AI may feel like a recent innovation, it’s actually been used by banks and financial institutions for a long time. The many different players in the financial services industry — from investment and retail banks, to insurance companies, to infrastructure providers like exchanges — all generate lots of data.
“This is why finance is so right for embedding AI into its operations,” says Ryan. “These market participants can very quickly and accurately analyze this data to make informed business decisions. Banks have really considered themselves to be technology companies operating in the financial services industry, and AI is another tool they can use to increase efficiency, optimize processes, and ultimately provide better service for their customers.”
Some ways AI is currently used in finance and banking:
Compliance: Fraud detection and ALM monitoring
Risk management: Alerting risk managers to potentially unacceptable risks
Consumer banking: Improve time- and cost-efficiency through chatbots
Wealth management: Helping wealth managers create tailored solutions for clients
Investment banking: Identifying companies that need to raise capital or are candidates for acquisitions
Trading: Algorithmic trading strategies or generating signals through sentiment analysis
So a tool like ChatGPT can, say, elevate the personalization and experience of digital banking through smarter chatbots, or help wealth managers instantly search massive databases of information to create more bespoke solutions for their clients.
“Actually EQT,” Ryan continues, “a European company under the third largest private equity company in the world, had their own AI tool that would analyze data and make suggestions about which companies might need to raise capital. When ChatGPT 4 came out, they integrated it with their in-house tool to create a much more powerful platform to help them with deal making.”
One particularly interesting area of AI is sentiment analysis, which is essentially processing and analyzing what people are talking about online. In trading, artificial intelligence tools will scrape data from social media and financial news platforms like Bloomberg to see what people are saying and thinking and use that to predict trends in the market or movements in a stock.
“Ive seen these tools get more sophisticated. You have these little AI plug-ins now and you can see some of the sentiment analysis to generate these scores, if you will,” says Joseph.
But what about emerging cases of artificial intelligence, beyond existing ones? Before getting into the future of AI in finance and banking, we conducted a poll from over 1,300 webinar participants to gauge their feelings about the AI unknown. The vast majority felt excited, a few worried, and many a combination of both.
The Future of Artificial Intelligence in Finance and Banking
Two critical things to consider as we move forward are the ethics and governance of artificial intelligence.
How can a machine be biased? Since the input AI models take is controlled or created by humans, human bias is still factored in. We have to be very careful that the input sample data into large models is truly representative of the population that the output is going to be making decisions about.
“We need to be hyper-aware of the implications of making decisions based on the output of these models, particularly if we don’t have a good understanding of how they came to those decisions,” Joseph emphasizes.
The second point is governance, or the regulation of AI. As new tools are released to the market at an increasingly rapid pace, regulation can lag behind.
“Recently, an open letter was penned asking to pause the development of large-scale AI models. As of this morning, it’s reached 28,000 signatures including people like Andrew Yang and Steve Wozniak,” says Joseph. They’re not asking for a pause on AI development in general, but to take a step back from the race to deploy “ever more powerful digital minds that no one can predict, understand, or reliably control.”
Nobody wants to halt progress. The ultimate goal is to have more accurate, safe, interpretable, transparent, robust, aligned, trustworthy, and loyal models.
How Finance Professionals Can Stay Relevant in the World of Artificial Intelligence
“The reality is, AI is here and it’s not going anywhere,” says Ryan. “Think about what transferable skills we already possess that will allow us to differentiate ourselves, despite all of the AI tools out there.”
Things like the ability to work collaboratively, time management skills, and being able to communicate and work effectively with your team members — these are all transferable, human skills that employers highly value.
It’s also important to think about the skills you don’t have and what you need to develop to be really valuable in an AI world.
“First of all, it’s to become digitally literate. To be able to talk about and understand what machine learning is, what deep learning is, and at least be engaged in that conversation,” Ryan starts. Developing emotional intelligence, empathy and interpersonal skills, as well as critical thinking and problem-solving abilities, also go a long way.
“Finally, develop your sense of agility, adaptability, and resilience. As the pace of technology development increases, our ability to deal with change and the stress of an ever-changing work environment — if we can become resilient, that will help us.”