Enhancing transparency and reliability in generative AI through case-based reasoning and explainable AI
Year: 2026
Type: article
Abstract: Enhancing transparency and reliability in generative artificial intelligence remains a critical challenge due to the inherent opacity and stochastic nature of deep learning models. A hybrid framework integrating Case-Based Reasoning (CBR) and Explainable AI (XAI) is proposed to address these limitations by combining generative capabilities with structured reasoning and interpretable outputs. The model leverages advanced generative systems such as ChatGPT, augmented with a case retrieval and adaptation mechanism that grounds outputs in historical knowledge. Experimental results demonstrate significant improvements across multiple dimensions, including generation quality (LLM Score: 0.927), factual consistency (0.914), explanation faithfulness (0.902), and trust calibration (0.921). Statistical validation confirms the robustness and significance of these improvements, with strong effect sizes observed across diverse tasks. The framework also exhibits high adaptability and stability under noisy, ambiguous, and multi-hop scenarios. By integrating reasoning transparency with adaptive learning, the proposed approach enhances user trust and system reliability. These findings establish a novel direction for developing trustworthy generative AI systems that align with real-world requirements for interpretability, accountability, and consistent performance.
Source: Array
Author: Thacha Lawanna
Institution: Chiang Mai University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,680
OpenAlex ID: W7162211754