Explainable Artificial Intelligence untuk Deteksi Kecurangan Pangan: Meningkatkan Transparansi dan Kepercayaan Konsumen pada Agroindustri Modern

Abstract

The increasing complexity of global food supply chains has intensified the risk of food fraud, resulting in food safety concerns, economic losses, and declining consumer trust. Although Artificial Intelligence (AI) has demonstrated strong capabilities in anomaly detection, food authentication, and food safety risk prediction, most existing models operate as black boxes, limiting users’ understanding of the decision-making process. This study aims to analyze the role of Explainable Artificial Intelligence (XAI) in enhancing the transparency of food fraud detection systems and strengthening consumer trust in modern agro-industrial environments. A qualitative approach was employed through a systematic literature review of reputable scientific publications addressing AI, food safety, food authentication, supply chain traceability, and XAI. The findings indicate that XAI improves model interpretability by explaining the factors influencing predictions through methods such as SHAP and LIME, thereby enhancing accountability, validity, and user acceptance of AI-generated decisions. Beyond supporting food auditing and regulatory monitoring, XAI contributes to greater information transparency, which is essential for building consumer trust. This article argues that algorithmic transparency serves as a critical mechanism linking AI predictive performance with social legitimacy in digital food governance. The study offers a conceptual contribution to the development of more transparent, trustworthy, and sustainable food fraud detection systems.