Estimated reading time: 5 minutes
GenAI went mainstream barely a year ago. But it is already changing the world. How will it impact financial services? Does it have the power to personalise products, improve customer care and cut fraud? Let’s explore.
In 2014 Ben Bernanke applied for a loan. He was declined. This was a little strange. After all, Bernanke was the former chairman of the Federal Reserve with a net worth of up to $2.3 million.
The problem? Bernanke had just gone ‘freelance’. For the agencies that calculate the US’s credit scores, freelancers with no track record are a high credit risk. So the application was declined – even though as a freelancer Bernanke could charge $250,000 for a one hour speech.
The Bernanke story reveals the limits of a crude credit scoring system, which was created in the 1950s when jobs and lifestyles were very different from now. Clearly, the system fails a substantial number of applicants – and deprives lenders of substantial revenue in the process.
The solution is obvious: use more data to assess someone’s credit worthiness.Today, we have so much data. And thanks to a huge leap in technology we can now analyze this information instantly to make much better decisions.
The huge leap in technology is, of course, artificial intelligence. AI powers systems that can crunch through millions of data points, and look for patterns that reveal whether someone is a good or bad risk. Indeed, firms such as Upstart and Zest Finance are already making waves in the lending space thanks to AI and machine learning tech.
But credit is just one area in which AI is disrupting financial services. Others include:
• Improving authentication and on-boarding
• Detecting fraudulent activity
• Personalizing financial products
• Enhancing customer care
• Document reading
• Speeding up code development
Needless to say, the above applications are just scratching the surface of what is possible. Analysts expect innovative start-ups to find many novel applications of AI tools. It’s why the global AI in fintech market size is expected to reach $41.16 billion by 2030, according to a report by Grand View Research.
So let’s take a closer look at the potential impact of AI on banking and finance.
What is AI and machine learning?
Before diving into the topic, we should establish exactly what AI is. This is not an easy task. Even the experts disagree!
In simplest terms, AI is a general concept to describe software that mimics the ways humans think in order to perform complex tasks. These tasks could include physical movement/robotics, decision making, data analysis, facial recognition, language translation and more.
Machine learning is a more precise term. It comprises a subset of AI that focuses on the use of data and algorithms. Using complex math, ML systems can make classifications or predictions to uncover key insights buried in large data sets. And they can ‘self-learn’ to improve their performance over time. Stanford University defines machine learning as “the science of getting computers to act without being explicitly programmed”.
Why is there so much interest in AI and machine learning now?
This is a very simple question to answer… and the answer is: Chat GPT.
ChatGPT was launched late in 2022 and made a huge impact almost immediately. In fact, it
reached 100 million users just two months after release. By comparison, Instagram took more than two years to reach the same milestone.
ChatGPT captured the public imagination because it was one of the first high profile examples of generative AI. To put this into context, the first wave of AI was all about classification. In this phase, programmers would train computers to classify various types of input data: images, video, audio, language.
Generative AI goes a step further. Here, the computer studies that input data in order to produce new data. In the case of ChatGPT, the software can generate original articles and even jokes or poetry in response to simple text prompts. Another successful example of generative AI is Dall-E, which generates images in a similar way.
For casual users, it all seems like magic. And there’s now much public debate about the possible impact of generative AI on every aspect of society. Naturally, there are many concerns. But there is also a sense that generative AI can deliver substantial economic benefits.
Indeed, earlier this year (2023), Goldman Sachs predicted that generative AI could drive a seven percent increase in global GDP over a 10-year period. That equates to almost $7 trillion.
AI in finance: key applications of the technology
How AI can solve real challenges in financial services? Here are six promising applications.
• #1. User interface and on-boarding
Many banks are already using AI bots to guide customers through product on-boarding, and answer general account queries. Speech recognition can make these encounters even more productive. When people can speak rather than type their queries, they can get answers quicker. Speech recognition also encourages customers who are less confident with digital services.
• #2. Sentiment analysis
Is a customer angry? Is a sales prospect about to say yes? AI bots are becoming sophisticated to know the answers. It will soon be possible for agents to get useful assistance from robo-advisors.
• #3. Anomaly detection
AI systems are built to analyze large data sets and to detect patterns in what they read. This is the key to detecting anomalies such as fraudulent transactions, financial crime, spoofing and so on.
• #4. Personalized products and services
Personalization is the dream for financial service providers. Thanks to AI, banks are reaching the stage where they can tailor products for individual customers. AI systems can study historical customer journeys, peer interactions, risk preferences, behavioral data and more to create personalized services around investments, savings, loans and so on.
• #5. Document processing
In the bank ‘backend’, AI tools can speed up laborious office processes. Banks accumulate huge libraries of structured and unstructured data, in which is buried useful strategic information. Historically, employees have trawled this data to discover actionable insights. AI can do this much faster. And now, thanks to large language models, it can output written reports too.
• #6. Image recognition
Financial services companies frequently have to study images and videos. Think of the insurance company accessing a claim for example. Or the bank performing an identity check. These are all areas that AI can improve thanks to advances in image recognition.
How Thales is helping banks to sniff out fraudulent transactions
Regrettably, the fraudsters have stampeded through this door. According to JP Morgan’s annual payments survey, card-related fraud in the US rose by 10 percent in 2022. The total cost of this fraud was $1.6 billion.
The best way to defend against these attacks is to create authentication systems that spot the criminals. This is where AI can help. We spoke to Xavier Larduinat, Marketing Manager for Banking & Payment Innovation at Thales, about the latest developments in AI for banking.
#1: Is AI already having an impact in finance and banking?
Yes it is. I’ve heard ChatGPT described as a shockwave across the industry. But it’s just the start. When Meta came to our HQ, they said ChatGPT is like a toy when compared to what’s coming next. We’re already seeing the improvements in service bots, for example. BNP Paribas says its chat bot has already generated 750 million queries.
#2: Your focus is on AI in authentication. How does it work?
We have developed a IdCloud offer especially for banks, which combines device intelligence, IP intelligence, digital habits verification and behavioral biometrics. Our system uses AI to study an authentication request. It looks at data points like the context of the request, the IP address, typical habits, time of day, location and much more. It then studies risk signals to calculate a risk score and make a decision about the identity of the customer. If there’s any doubt, it asks the user to complete more steps.
#3: Have you rolled out these systems?
Yes. We are working with around 100 companies – helping them to improve their on-boarding processes and so on. We call it device intelligence AI, and we believe it can reduce suspicious authentication attempts by 90 percent.
#4: What do consumers think? Do these systems add more friction?
Most of the time, the systems reduce friction because they perform their checks in the background. But I think customers like it when they are asked for more proof of ID. It proves to them that the banks are taking care.
#5: Is there an ethical problem if an AI makes decisions for a bank, rather than a person?
I think everyone is aware of this challenge. Decisions can’t be made in a black box. Banks must be able to explain why they make the choices they do. But this is not the only issue. They have to ensure they have sovereignty over the training data too, for example. There must be regulation in this area.
At Thales we launched the TrUE approach, which stands for Transparent AI. We want to ensure that users can see the data used to arrive at a conclusion, and to follow objective standards protocols, laws, and human rights. We also joined the confiance.ai consortium, which is a body committed to explaining black box operations and guaranteeing that data mining is limiting to the purpose of risk management.
#6: How else can AI impact financial services?
Lots of ways. Customer care is an obvious area. There are already bots and robo-advisors out there, and I think AI will make these services even more useful. For example, they will get better at understanding sentiment. But I also believe AI can bring other benefits. By speeding up the writing of code and making processes more efficient, it can help banks to reduce their environmental impact.
And AI could help financial inclusion too. Digital financial services can be complex. If AI contextual analysis can remove steps from the user experience, it will help more people to access these services.
Experts agree that the best way to think about AI is to see it as a ‘super assistant’ – one that that can help people and organizations to complete tasks faster and more accurately. In finance, with its huge datasets and complex regulatory requirements, the tech is a perfect fit. Over the coming years, it promises to change the way finance companies work, and transform the way customers manage their money.
How traditional banks can counter neo-banks and fintechs with digital-first mobile services
How to make central bank digital currencies work for offline payments
On the money? Seven key banking and fintech trends for 2023