Using ChatGPT and other LLMs in Professional Environments

Abstract

Large language models like ChatGPT, Google’s Bard, and Microsoft’s new Bing, to name a few, are developing rapidly in recent years, becoming very popular in different environments, and supporting a wide range of tasks. A deep look into their outcomes reveals several limitations and challenges that can be further improved. The main challenge of these models is the possibility of generating biased or inaccurate results, since these models rely on large amounts of data with no access to unpublic information. Moreover, these language models need to be properly monitored and trained to prevent generating inappropriate or offensive content and to ensure that they are used ethically and safely. This study investigates the use of ChatGPT and other large language models such as Blender, and BERT in professional environments. It has been found that none of the large language models, including ChatGPT, have been used in unstructured dialogues. Moreover, involving the models in professional environments requires extensive training and monitoring by domain professionals or fine-tuning through API.