SMU Data Science Review


The advancement of large language models (LLMs) has significantly expanded the influence of artificial intelligence across various sectors. This paper explores building LLM agents to power applications and examines what is necessary to build an efficient and helpful AI assistant. The research investigates the core components necessary to create specialized agents, facilitate collaboration in problem-solving, and improve human task performance. The development and application of tools designed to augment the capabilities of LLM agents are also explored. The paper addresses the potential risks of the unknowns, such as hallucinations, which can compromise the success of agent-based solutions within LLM applications. The potential application areas for LLM agents and the broader implications of these findings on AI research and application development are discussed.

Creative Commons License

Creative Commons Attribution-Noncommercial 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License

Included in

Data Science Commons