A.I. speaks about AI at Risk Live Europe 2024

Leveraging AI & LLMs L-R
Zoi Fletcher, Kilhan Shah, Andrew Mackay, Andreas Ita
Risk Live Europe  
Photo: Tom Parkes
©Tom Parkes
+44 (0)7831214060
  • Posted on June 25, 2024 by Dr. Andreas Ita

At last week’s Risk Live Europe 2024 in London, Orbit36 Managing Partner Andreas Ita (A.I.) was invited to join a panel discussion about the use of Generative Artificial Intelligence (Gen AI) and Large Language Models (LLM) in Risk Management.

Since we note increasing interest in such technologies from our clients, prospects and industry peers, here a few thoughts on this important subject:

Evidently, AI technologies can provide expanded analytical power, speed and at the same time significant cost saving opportunities.

Since more than two decades, large global banks started to outsource and/or offshore a significant part of their labor-intense back-office, finance and risk management tasks into offshoring centers in Asia and Eastern Europe. At the beginning, the individual tasks and process steps were not changed much, but simply executed by cheaper workforce. AI provides now the logical continuation of this trend. With automated tools replacing the manual work, executing processes becomes cheaper again. Further, efficiency gains through AI are also accessible for smaller and mid-sized banks, which so far could not benefit from labor outsourcing. We believe that AI could become a game changer in the banking industry, as it permits also smaller institutions to capture cost advantages. Economies of scales could therefore become less important for the banking sector than in the past.

Solely relying on AI to achieve quick cost savings will though be insufficient. We believe that many processes in banks are unnecessary complex and cumbersome for clients. We regard it as equally important for banks to invest into well-defined, lean processes, which improve customer experience. Rethinking processes and designing them in a way which can be easily automated bears potential which goes beyond simple cost savings. We therefore think that banks need to enhance and simplify their processes in parallel to the use of AI technologies.

When it comes to risk modelling and regulatory reporting, Gen AI and LLM could be less powerful than commonly believed. Since regulations are often prescriptive and results need to be replicable, such processes will likely continue to be automated in a traditional way. The necessity to document and independently validate models provides a further restriction.

LLMs may, however, fundamentally change the way how risk management is performed. Today, executives and board members have largely to rely on reports if they want to understand the risks in their firm. As recent incidents have shown, risk and finance reports provide sometimes an incomplete or outdated picture of the real situation. For instance in the default of Archegos, it was critical for the decision makers to quickly understand the true magnitude of their exposures. We believe that LLMs bear untapped potential to provide managers in critical situations the information they need. Too often the necessary data would have been available within the bank, but could not be collected and prepared timely enough. LLMs could do an excellent job in providing risk managers the information they need in real-time, even when stored on different systems. Also the ongoing monitoring of client behavior will likely become more effective, possibly allowing banks to detect fraudulent pattern already at an early stage.

Orbit36 is following the further development of new technologies with interest, since it will likely change the way how risks are managed in banks already in the near future. We see a lot of potential to complement existing risk management methods by very powerful new instruments. Especially for dynamic risk types like market risk or counterparty credit risk, real-time risk management could become the new reality.

Please approach us if you would like to discuss the opportunities provided to your firm.

    Leave a Comment

    Your email address will not be published. Required fields are marked *

    Scroll to Top