Generative Artificial Intelligence for Industry: Opportunities, Challenges, and Impact

Abstract

The recent advances in Generative Artificial In-telligence (GenAI) and Large Language Models (LLMs) have generated significant interest across the world. For a successful adoption of GenAI and LLMs by industry, it is critical to identify their potential benefits, impact, and challenges. Accordingly, in this work, we investigate a few use cases of LLMs, which are relevant across most industry segments. In order to empirically evaluate the impact of GenAI on the code generation use case, we build CodePrompt, a handcrafted dataset of sequential prompts used by a human user to generate code. We approximate efficiency by considering the ratio of the number of tokens of code generated by an LLM to the number of tokens in the user's prompt. Experimental results reveal that a sequential trial of prompts for code generation may lead to an efficiency factor of about 6.33, on average, which means that a user's effort is reduced to about one-sixth.