Enterprise large language models: Knowledge characteristics,risks, and organizational activities |
| |
Authors: | Daniel E O'Leary |
| |
Institution: | Marshall School of Business, University of Southern California, Los Angeles, California, USA |
| |
Abstract: | Since the release of OpenAI's ChatGPT, there has been substantial interest in and concern about generative AI systems. This paper investigates some of the characteristics, risks, and limitations with the enterprise use of enterprise large language models. In so doing, we study the organizational impact, continuing a long line of research on that topic. This paper examines the impact on expertise, the organizational implications of multiple correlated but different responses to the same query, the potential concerns associated with sensitive information and intellectual property, and some applications that likely would not be appropriate for large language models. We also investigate the possibility of agents potentially manipulating the content in these large language models for their own benefit. Finally, we investigate the emerging phenomenon of “ChatBot Enterprise” versions, including some of the implications and concerns of such enterprise large language models. |
| |
Keywords: | BARD BlenderBot ChatGPT ChatGPT Business ChatGPT Enterprise enterprise generative AI enterprise large language model (ELLM) experimental analysis expertise generative AI manipulation generative AI systems (GAIS) human-in-the-loop large language models (LLM) |
|
|