Artificial intelligence (AI) is shaping digital transformation – even in the financial sector. Ever larger volumes of data and more efficient processing power are key factors in this. But what’s the real advantage of AI over tried-and-tested IT applications? “Before AI, every step had to be programmed,” explains Lukas Bütikofer, a data scientist at PostFinance. Even the slightest deviation from the standard could cause systems to stop working. “AI, on the other hand, learns to classify data independently using known cases and can also be trained for complex tasks.”
Artificial intelligence at work
Artificial intelligence is a major driver of digitization. Two members of the data science team explain the growing role of AI at PostFinance.
Weak and strong AI?
When talking about artificial intelligence, a distinction should be made between “weak” and “strong” AI. “Weak AI means the AI system is developed to solve a specific task,” says Nicole Keller from the Data Science Team. In comparison to strong AI, which has human-like cognitive abilities, the weak system has no deeper understanding of problem-solving. What this strong AI – also called “artificial general intelligence” – will one day be capable of remains to be seen. However, weak AI is already in use in many areas. Here are three examples:
Area 1: positive customer experience
PostFinance uses AI to optimize customer experience. For one thing, analyses help predict customer requirements. Then virtual advisors in the form of voice assistants or chatbots are used to improve customer satisfaction. “Our digital assistant recognizes customer queries and tries to respond to them directly,” says Nicole Keller. The biggest challenge is to accurately understand the customer’s needs during initial contact. The digital assistant already does a very good job of answering numerous FAQs and so should be able to process customer data in future. “Authenticated customers will, for instance, also be able to check their account balance via the assistant.” Keller believes personalized advice on financial products or queries on share prices would also be within the realms of possibility.
Area 2: improved risk management
Financial risks can also be better calculated with AI. Analysing large volumes of unstructured data produces sounder analyses and more accurate forecasts. This in turn improves returns on funds, for example, and reduces the risk of default on credit. “When lending, information on thousands of previous credit decisions and customers’ repayment behaviour is helpful,” explains Lukas Bütikofer. The AI system compares recognized patterns against the current application, and in this way assesses creditworthiness. “However, this is only as good as the existing data and therefore also entails certain risks.” For this reason, the decision should not rest on this type of system alone.
Area 3: optimized fraud detection
AI is playing an increasingly important role behind the scenes at PostFinance, too. One example is in the detection of fraud. AI applications monitor cash flows, assisting in the early detection of payment discrepancies and helping to prevent money laundering. “Some tasks cannot be performed quickly enough manually,” points out Lukas Bütikofer. “So here I’m thinking of fraudulent transactions, for instance.” An AI system learns to distinguish between normal and suspicious transactions and flags up possible cases of fraud. This enables prompt action to be taken. “Other tasks are so complex that we humans are either unable to solve the problem, or AI achieves far greater accuracy.” One such example given by our expert is the detection of unauthorized login attempts in e-banking on the basis of behavioural patterns such as typing speed.
According to Nicole Keller, the current priority in AI is integrating the technology into existing processes in already identified areas and using it productively. The exciting challenges here include issues such as cybersecurity, privacy, and data protection, as well as data-related risks such as customer discrimination or reliance on large technology providers. Another key issue is the integration of new applications into existing systems, as it’s not possible to simply replace these systems with new solutions. “This is a challenge for us. We have to find ways to augment the traditional systems with the new technologies.”
The future belongs to artificial intelligence
Nicole Keller and Lukas Bütikofer believe there’s great potential in AI. From a technical point of view especially, the limits are almost boundless. They firmly believe there will be big advances both at PostFinance and in the industry as a whole over the next few years. “A lot of that will go beyond what we can even imagine right now,” remarks Nicole Keller. Lukas Bütikofer adds: “At any rate, thanks to new products and applications everything will be easier, more intuitive, faster, and also less expensive.”
Would you like to help shape the digital transformation?
About Nicole Keller and Lukas Bütikofer
Nicole Keller has worked at PostFinance since 2014. She holds a BA in International Relations, an MSc in International Business Development, and an MASt in Data Science. Asked why she enjoys working as a data science specialist at PostFinance, she tells us: “Because here I’m part of the best data science team in the world, and what I do has an impact on millions of customers.”
Lukas Bütikofer has been with PostFinance since 2017. He has a master’s degree and a PhD in Physics. When asked why he enjoys working as a data science specialist at PostFinance, he tells us: “At PostFinance, I see a great deal of potential and the opportunity to help shape the future of this field.”