KubroTM > Takeaways from the LLM in Investment Management panel
Takeaways from the LLM in Investment Management panel
Posted on 10 November, 2023

A few takeaways from our panel discussion on the applications of Large Language Models and AI in investment management at the 5th Annual Beryl Elites conference in NY the other day, with Peng Cheng (JP Morgan), Gareth W Shepherd (Voya Investment Management), Judith Gu (Scotiabank), and Chris Brown (NVIDIA):

💡 A lot of excitment about the LLMs but still mainly at an experimentation / POC stage, a lot more happening in broader AI applications

💡 Fine-tuning open source models with proprietary data is what many would prefer, but all kinds of experiments and approaches are under way

💡 Causality / causal inference now a hot topic, with methods to get more scientific in finance ('randomized controlled trials in finance' - ways to approximate that)

💡 Important to distinguish between decision-oriented applications (signals) vs operational efficiency (automation)

💡 Humans still key, incl. required expertise to deploy tech; focus on augmenting the human intelligence

💡 Investment management sees much less 'LLM/AI action' than some other sectors like healthcare and media

Thanks to Vidak Radonjic and the Beryl team for a great event.