Exploration of DeepSeek's Application in Auditing—Based on Tests of Ernie Bot, DeepSeek, Kimi, and Doubao
The Frontiers of Society, Science and Technology · 2026 · 8 (2)
DOI: 10.25236/fsst.2026.080204 ↗Kurz gefasst
Fachartikel in der Zeitschrift The Frontiers of Society, Science and Technology, 2026.
Abstract (Originalsprache)
: The DeepSeek large language model (LLM) brings transformative opportunities to the auditing industry with its advantages of high efficiency and low cost, multimodal processing, professional knowledge base, and high-precision analysis. To explore its real-world performance and application prospects in the auditing field, this study conducted a comprehensive evaluation of general-purpose domestic LLMs represented by Ernie Bot, Kimi, and Doubao, as well as the high-performance open-source inference model DeepSeek, using an examination-based testing method across three rounds of quantitative tests. The test data included real questions from the Junior Auditor Qualification Examination. From the results, it is apparent that LLMs such as Ernie Bot, Kimi, and Doubao excel at answering memorization-based questions but have some deficiencies when addressing comprehension-based questions, and thus cannot wholly replace professional auditors. DeepSeek, despite failing to replace experienced auditors, excels in memorization-based questions and has great advantages in answering questions that require critical thinking and deep understanding because of its superior logical capability, fast inference speed, economical training and inference, and outstanding code generating capability. In light of the results presented in this research, we propose an approach for incorporating DeepSeek into auditing practices to take advantage of its capabilities, relieve auditors from excessive workload, improve audit quality and efficiency, and offer valuable insights into the auditing sector in the age of AI.
Zitieren
APA
(2026). Exploration of DeepSeek's Application in Auditing—Based on Tests of Ernie Bot, DeepSeek, Kimi, and Doubao. The Frontiers of Society, Science and Technology, 8(2). https://doi.org/10.25236/fsst.2026.080204
Vancouver
Exploration of DeepSeek's Application in Auditing—Based on Tests of Ernie Bot, DeepSeek, Kimi, and Doubao. The Frontiers of Society, Science and Technology. 2026;8(2). doi:10.25236/fsst.2026.080204
BibTeX
@article{euvimed2026,
title = {Exploration of DeepSeek's Application in Auditing—Based on Tests of Ernie Bot, DeepSeek, Kimi, and Doubao},
journal = {The Frontiers of Society, Science and Technology},
year = {2026},
volume = {8},
number = {2},
doi = {10.25236/fsst.2026.080204}
}
EUVIMED