How banks use ChatGPT and artificial intelligence
With the release of ChatGPT by OpenAI at the end of November 2022, a new chapter in the history of chatbots based on artificial intelligence (AI) opened. ChatGPT can understand human input, analyze, summarize, translate, and rewrite text, program websites and Excel macros, or even correct faulty software code. The far-reaching possible applications have also led the financial industry to think about possible use cases.
Where banks can save money thanks to AI
Probably the most obvious area of application for language models and chat programs is in the interaction with customers. Instead of the mostly marginally helpful chatbots with a limited repertoire of possible answers, a program like ChatGPT, which is based on a huge amount of data and works with natural language processing (NLP), could handle customer inquiries and complaints in a satisfactory manner around the clock in real time.
According to a study by Signicat, more than a third of Gen Z respondents feel their bank onboarding process is taking longer than expected. About a quarter of those surveyed abandon the process because it is too lengthy or detailed. It is estimated that banks lose over 5 billion euros annually due to aborted onboardings. With the help of AIs, the digital onboarding of new customers can be made more intuitive: Instead of having to fill out endless forms when opening an account, in future you could simply be guided through the onboarding process as part of an online conversation with a chatbot — with the possibility of an immediate (polite) answer if you have a question.
In the field of asset management, robo advisors have long since established themselves on the market. Global assets under management through robo advisors are expected to nearly double from approximately $2.45 trillion in 2022 by 2027. It is estimated that more than 200 million users currently rely on the services of a robo advisor. The willingness of customers to rely on algorithms and artificial intelligence to manage their assets is therefore quite high. It therefore seems only logical to use language models such as ChatGPT to make the interaction with these customers more natural and individualized. In this way, current financial market developments, changes in the portfolio, new investment proposals or questions about the investment strategy can be prepared and communicated in a customer-oriented manner.
In addition, thanks to AI, challenges in the area of accessibility can also be addressed by digital tools. The proportion of people aged 80 or over almost doubled within the EU between 2001 and 2021. The proportion of people over 65 has increased in all Member States over the same period. Not only since the media campaign “I’m old, not stupid” by a Spanish pensioner, banks have to deal with how to make their digital services accessible to all sections of the population. For example, how should a mobile app for older people be designed? With automatic speech recognition (speech to text) and voice output, people with visual and physical disabilities could, for example, query their account balance or have the AI record a transfer to their savings account.
There are also numerous back-office activities that can be improved, simplified, and accelerated by AI. With the ability of AIs to quickly analyze large amounts of data and identify patterns and anomalies in that data, they can significantly support bank’s back-office staff in their tasks.
For example, AIs are used to analyze transaction data to identify possible criminal payment flows (keyword: AML) or suspicious transactions — in real time. The ability to review transactions as they are cleared and make informed decisions immediately is gaining in importance due to the proliferation of real-time payment systems such as SEPA Instant Credit Transfers. By integrating a wide variety of data sources, AIs can also be used in a bank’s holistic risk management. In addition to internal data, for example, external data on economic and political developments can also be analyzed by the AI and a corresponding assessment of the resulting risks for the bank can be issued. As a knowledge carrier within a bank, a chatbot based on Large Language Model (LLM) technology can also provide quick and uncomplicated help with inquiries from employees. This can also simplify the onboarding of new employees, for example.
With the introduction of Microsoft 365 Copilot, ChatGPT will find its way into many workplaces in the medium term. The aim of the feature is to reduce the workload for non-essential activities through functions such as summarizing meetings, e-mails, and chat histories, to increase the productivity of employees and to concentrate resources on value-adding activities in the sense of lean management. Banks can not only save costs, but also position themselves as more attractive employers with new and interesting areas of responsibility.
Risks of using AI
With AI tools like ChatGPT, the data entered by users is used to further train the AI. Some investment banks, such as Deutsche Bank AG and Goldman Sachs Group Inc., have therefore banned their employees from using AI tools. The background is the fear that entered, business-relevant or confidential data will thereby reach external parties.
It is clear that AI tools, like other software, must be tested before implementation. In Europe, the EBA guidelines for the management of ICT (information and communication technology) and security risks as well as the regulatory specifications, e.g. by the FMA in Austria or the BaFin in Germany, contain some corresponding specifications for banks and financial service providers. In the case of outsourcing ICT services and systems to third parties, financial institutions must ensure that information security is guaranteed. Especially in connection with data protection (keyword: EU-GDPR), this also includes the life cycle of the data, data encryption and the location of the data center.
Despite the great advances in chatbot development, the systems are not flawless. For example, the ChatGPT database only extends to the year 2021 and abstract questions can sometimes lead to wrong answers. In a sensitive area like the financial industry, this can quickly cause significant problems. Accordingly, human control is mandatory both when training the AI and when continuously developing it by expanding the data sets.
Netcetera as a pioneer and long-term partner of banks worldwide
With its products and services in the areas of secure payments, digital banking, and digital identity, Netcetera enables banks to use the advantages of digitization holistically. Thanks to our many years of experience in the financial industry, we recognize trends at an early stage and provide our customers with solutions that meet both current and future market requirements. Would you like to learn how to implement your digital vision? Then contact us or visit us at one of our locations.
