| 摘要: | 人工智慧（AI）技術（如科技金融(fintech)，機器人投資顧問(robo-advisor)，算法交易軟件(algorithmic trading software)和區塊鏈(blockchain)）已逐漸應用於全世界各地的公司，內部稽核可被視為企業內部控制和風險管理的守護者，人工智慧應用在內部稽核過程中，可以提高效率和責任，確保實現內部控制目標，促進企業的長期穩定發展。大數據和雲端運算背景下的財務報表數據己經常常聚集在企業中，因此全球四大會計師事務所皆將人工智慧技術導入會計師事務所的審計程序中，會計師對於設有內部稽核的企業皆會評估其內部稽核品質，作為會計師評估審計客戶內部控制的參考，而內部控制的瞭解與評估是會計師執行審計的必要程序。因此本計劃在探討人工智慧技術下企業內部稽核架構之評估與增進後，將再進一步探討人工智慧下審計程序之評估與增進。人工智慧應用在內部稽核及審計程序是屬於多目標決策問題(MRDM)，本研究計劃採用一種混合決策架構，此架構結合軟體計算技術(蟻群基礎的模糊約略集合理論 (ACO-FRST) 及變數一致之優勢關係約略集合理論（VC-DRSA）及MADM，用以檢測實現最佳期望水平的績效值和差距缺口，並提出優先改善的策略方案。幫助企業在其內部稽核程序中有系統地增進人工智慧應用程序，以防止企業內部控制風險暴露和會計師事務所的審計失敗。
Artificial intelligence (AI) technologies (such as fintech, robo-advisor, algorithmic trading software, and blockchain) have been gradually applied to companies around the world. Internal auditors can be viewed as the guardians of an enterprise’s internal control and risk management. As such, the application of AI in the internal auditing process can improve efficiency and accountability, ensure the achievement of internal control objectives, and promote the long-term stable development of the enterprise. The financial statement data in the context of big data and cloud computing has often gathered in enterprises, so Big 4 CPA firms have introduced artificial intelligence technology into the auditing process. CPA will assess client’s internal auditing quality, as an accountant to assess the internal control of audit clients, and the understanding and evaluation of internal controls is the necessary procedure for CPA auditing process. Therefore, after executing the evaluation and promotion of the AI internal auditing structure, our plan will further explore the evaluation and enhancement of the AI auditing process. Strategies related to AI applications in enterprise internal auditing and CPA auditing process evaluation and improvement planning are inherently multiple-rule/rough-based decision making (MRDM) issues and are essential to the stability of corporate and CPA firms operations so as to reach the goal of sustainable development. To overcome these tasks, this study introduces a hybrid decision-making framework that integrates soft computing technique (i.e., Ant Colony Optimization Based-Fuzzy Rough Set Theory, ACO-FRST, Variable-Consistency Dominance-Based Rough Set, VC-DRSA), and multi-attribute decision making (MADM) algorithm (i.e., DEMATEL, ANP and modified-VIKOR). The DEMATEL approach is used to analyze the influential influence relation map (IIRM) between the criteria and dimensions of AI internal auditing and CPA auditing. DANP (DEMATEL-based ANP) is then employed to calculate the influential weights of the dimensions and criteria. Finally, the modified VIKOR method is utilized to provide improvement priorities of performance AI internal auditing and CPA auditing. Internal and external auditors/users can thus consider the potential implications of our research findings to formulate better future policies that assist corporates in systematically improving AI applications in corporate internal auditing procedures and CPA auditing process as well as to prevent risk exposure and audit failures. |
