International Journal of Science and Research (IJSR)

International Journal of Science and Research (IJSR)
Call for Papers | Fully Refereed | Open Access | Double Blind Peer Reviewed

ISSN: 2319-7064


Downloads: 4 | Views: 146 | Weekly Hits: ⮙1 | Monthly Hits: ⮙1

Research Paper | Management | United Kingdom | Volume 13 Issue 1, January 2024 | Popularity: 4.7 / 10


     

Shaping the Future of Tax Advisory Using Artificial Intelligence

Conor Kelly


Abstract: This research harnessed the capabilities of GPT-4 using Langchain and Vector Search to develop an automated Tax Advisory Service, termed TaxBot. In light of recent advancements in Large Language Models, the project aimed to automate facets of the tax advisory domain, an industry traditionally reliant on skilled labour. TaxBot exhibited notable accuracy, achieving 70% on Charted Accountant Proficiency Level 2 Examination questions, and even up to 80% in specific tax areas like VAT. Efforts were made to enhance its reliability by labelling chunk embeddings, although challenges related to consistency persist. The project underscores the potential of AI in reshaping the tax advisory landscape, indicating a promising trajectory for the integration of AI technologies in professional advisory services. Future work will focus on refining the system for real-world commercial applications.


Keywords: Artificial Intelligence in Tax Advisory, GPT-4 Application, Retrieval Augmented Generation, Vector Search, Large Language Models, Automated Tax Services, Accountant Proficiency Examination, Value-Added Tax (VAT) Automation, Chunk Embeddings in AI


Edition: Volume 13 Issue 1, January 2024


Pages: 1381 - 1392


DOI: https://www.doi.org/10.21275/ES24115141827



Make Sure to Disable the Pop-Up Blocker of Web Browser




Text copied to Clipboard!
Conor Kelly, "Shaping the Future of Tax Advisory Using Artificial Intelligence", International Journal of Science and Research (IJSR), Volume 13 Issue 1, January 2024, pp. 1381-1392, https://www.ijsr.net/getabstract.php?paperid=ES24115141827, DOI: https://www.doi.org/10.21275/ES24115141827