Unstructured data hold a wealth of potential for both individual institutions and the financial system as a whole, says Bundesbank Executive Board member Joachim Wuermeling. “The upcoming developments in the worlds of AI and finance are presenting us with incredible opportunities, he said at BaFinTech 2023, an industry conference co-hosted by the Bundesbank and BaFin in Berlin. Leveraging the data in a way that maximises the benefits and minimises the risks is what counts, he remarked. He also cautioned that an appropriate balance needs to be struck in the regulation of artificial intelligence (AI). It would be unwise to curb the immense potential for innovation offered by these technologies by imposing excessive legal restrictions.

Mr Wuermeling explained that unstructured data are neither clearly organised nor clearly defined. They include multimedia content such as podcasts, videos, audio files, text messages, emails and the like. This distinguishes them from clearly structured data such as stock exchange prices, which are precise and systematic. Around 80% of the data collected by enterprises are unstructured, Mr Wuermeling noted, adding that this volume amounted to an astonishing 82 zettabytes – a zettabyte being a number followed by 21 zeros – worldwide in 2022. 

Data unlock key insights for enterprises

Enterprises can harness targeted data analytics driven by AI and machine learning to gain vital insights for their forecasts, risk assessments and decisions. According to Mr Wuermeling, leveraging these data is crucially important for banks, not just technically but strategically as well. Targeted unstructured data analytics helps propel cost efficiencies, automation, fraud prevention and improved risk management. Data-driven analytical approaches enable banks to make more precise decisions in their credit assessments or asset management operations. That improves resilience to fraud or cyberattacks, leaving banks more stable overall, Mr Wuermeling argued. Unstructured data analysis in the insurance industry has already delivered a number of benefits, including fewer cases of fraud and improved claims assessments. 

Mr Wuermeling also used his speech to showcase some examples of how the Bundesbank is leveraging unstructured data. A “BUBA-Bot” is helping employees in the Directorate General Markets gather valuable information on financial markets. A ChatGPT AI is currently being integrated into this tool. Furthermore, the Bank’s Directorate General Banking and Financial Supervision is developing a prototype that aims to harness unstructured data analysis to identify developments in bank risks and emerging trends. Because central banks have access to valuable market and supervisory data, they can leverage the benefits that AI has to offer and, in so doing, do an even better job of fulfilling their price and financial stability mandate, he said.

Using and understanding data properly

However, these innovations also harbour risks. There’s a danger that incomplete or faulty data might deliver unreliable results, Mr Wuermeling warned, adding it is risky to rely on data assessments without in-depth expertise. Owning data is not enough – using and understanding them properly matters as well, he cautioned. The general public’s perception of AI is another risk that should not be taken lightly. According to a survey by the Allensbach Institute, 58% of the general public dislike the term “artificial intelligence”. Acceptance issues of this nature could significantly impede the rollout of this technology, especially in the finance sector where trust is key, Mr Wuermeling argued. 

The BaFinTech industry conference is a regular event for representatives from supervised and non-supervised entities in the financial industry – in particular fintech firms, credit institutions, insurers and asset managers. The target audience also includes employees from software and hardware developers, technology service providers and experts from academia and industry.