LLMs generate text creatively but also adhere to literal or factual information when prompted. By mixing creative variability with literal information, current on-the-market LLMs manipulate us into applying theory of mind to things that don’t have one. This manipulation is decisively to their advantage, but probably not the user’s.
The Perilous Complexity of Information Extraction from Financial Contracts
March 7, 2021, 3:38 p.m.
A stark repeating pattern from my years of applying NLP to financial services is this: most seemingly straightforward information extraction tasks lure you with the sweet siren song of simplicity, allowing you to get just comfortable enough before the bottom falls out.