In this wide-ranging interview, Dr. Christoph Nieuwoudt, Chief Data and Analytics Officer at FNB, speaks to ITWeb News Editor Admire Moyo about how the Big Four banks are innovating with generative AI. Dr Newart also spoke about FNB's plans to hire more data analysts from the bank's graduate programs, with the aim of building solutions from emerging technologies, as the bank increasingly leverages generative AI. #itwebtv #fnb #genai
First National Bank (FNB) is building a vector database to support the generative artificial intelligence (AI) models it uses in its efforts to improve the functionality of its mobile applications.
Vector databases enable generative AI (GenAI) by not only providing efficient storage, retrieval, and representation of data, but also by enabling the transfer of learning and data augmentation techniques that improve the performance and robustness of generative models. plays an important role in supporting.
To build this vector database, the big four banks employ graduates from across South African universities to train their systems to efficiently respond to banking inquiries.
So said Dr. Christoph Neuwort, Chief Data and Analytics Officer at FNB, in a wide-ranging interview with ITWeb TV. He also revealed that the bank is considering deploying GenAI-based agents to answer various customer questions.
In 2022, Nieuwout told ITWeb the bank is looking to build a corps of 1,700 data analysts.
“Right now, we have over 2,000 full-time and contract data analysts, an area that is rapidly growing from around 1,300 probably two or three years ago. It continues to evolve and GenAI This is the latest topic to promote it.”
GenAI has the potential to revolutionize many aspects of banking, from customer service and risk management to marketing and cybersecurity, helping banks increase efficiency, reduce risk, and deliver more personalized services. He points out that they will be able to offer it to their customers.
“Although we are not growing as fast as we would like, we are always hiring new graduates. One of the main reasons is that there has been a significant change in the skills needed.”
Nieuwoudt points out that GenAI means a variety of skills are needed. For example, he explains that if you want to use large-scale language models or his GenAI library, you need to program in Python.
He said the reason for building the vector database is because banks are not using public GenAI models such as ChatGPT, which are not secure.
“If you think about ChatGPT, it's a public bot engine that people can use. [ChatGPT] Using Bard or other public bots does not guarantee the privacy of your information,” he said.
“As an organization, we cannot allow our staff to use these tools for banking purposes. Private virtual clouds have systems that can use similar underlying large-scale language models, but The system is placed in a ring-fenced environment, and all data fed into it and questions from users are placed there.
Dr. Christoph Neuwort, FNB Chief Data and Analytics Officer, said: (Photo by Leslie Moyo)
This means that FNB can fully protect the privacy of our clients' data. “We can ask ChatGPT about any topic in the world, but that doesn't necessarily mean it will be helpful to us. At banks, when we ask a bank to do a specific task, we guide them and prompt engineering. Must be able to.
“We are building a vector database, which is internal content that we look at to make sure our answers are accurate for our purposes.
“This is exactly why we need new skills. [the vector database] up requires specialized skills so that other staff can use it to understand the limitations and learn how to phrase questions.
“People often tell me about these systems. [GenAI] They were really stupid and said, “I asked this question and look at the answer.” But before looking at the answer, we need to look at the question. They're not as smart as humans, so have you given them the context to really understand what they need to respond to? Understand who they're talking to, what their role is, and what they're interested in. I have not. I don't understand any of this, so I need to give it more context. ”
Nieuwoudt said the bank aims to leverage data analysts to provide a personalized experience and improve search and chat features on the app.
“Today, we find that the FNB app is incredibly feature-rich. We often say internally that the app has too many features. There are over 100 different applets to choose from, but However, many of our customers don't even use a fraction of its features. We want to make it easier for our customers to ask questions if they want to.
“We currently use natural language to answer questions, but we are not yet able to provide a great answer that summarizes the answer and provides a link.
“We already have a secure chat feature where you can talk to your banker, but the banker doesn’t respond right away. I'm trying to route to an agent.
He added that the bank has direct partnerships with four universities, but hires graduates from all universities across the country.
“What we're looking at is people who are typically doing a four-year degree. So if you're doing a three-year degree, you should also be doing an honors degree. People with a master's degree are also interested, but they don't necessarily have to have a Ph.D., as only a few people have one.
“My field of expertise typically looks at mathematics and statistics, as well as engineering, actuarial science, and computer science.”
The team is interdisciplinary, he says. “He doesn't need five people with the same skills. Usually he needs a data scientist, a data engineer, a program manager, etc.”
