Hybrid Intelligence in Blockchain: Toward a Unified Model of AI-Enhanced Smart Contracts

Abstract Blockchain technology and artificial intelligence (AI) were introduced and integrated to develop new possibilities for security, automation, and transparency in digital financial ecosystems. This study proposes a hybrid AI Blockchain framework for automated credit evaluation and identity tokenization, addressing issues with opacity, bias, and data manipulation that affect conventional, centralized credit scoring systems. To predict client creditworthiness from past financial behavior, we trained several supervised machine learning models on the UCI Default of Credit Card Clients dataset. These models included Logistic Regression, Decision Tree, Random Forest, XGBoost, SVM, KNN, and Gradient Boosting. Support Vector Machine (SVM) outperformed the others in accuracy, precision, recall, and F1-score. Their performance metrics ranged between 0.841–0.915 for accuracy, 0.830–0.902 for precision, 0.828–0.900 for recall, and 0.829–0.901 for F1-score. Among all models evaluated, the Random Forest (RF) classifier performed best, achieving an accuracy of 0.907, a precision of 0.895, a recall of 0.894, and an F1-score of 0.895. Ensemble methods, however, achieved higher scores overall; among all seven classifiers, XGBoost obtained the highest raw accuracy (0.915), while Random Forest (RF) was selected as the deployed model, achieving an accuracy of 0.907, a precision of 0.895, a recall of 0.894, and an F1-score of 0.895. Random Forest was preferred over both SVM and XGBoost for production integration because it combined near-top predictive performance with substantially lower training/inference latency, native feature-importance interpretability, and simpler on-chain integration properties that matter for a system whose decisions must be auditable and re-verifiable. RF demonstrated an excellent balance of predictive accuracy, generalization, robustness to overfitting, and interpretability, making it the optimal choice for deployment in the blockchain ecosystem. The Random Forest model’s decisions were integrated into an Ethereum-based smart contract, which issues non-transferable identity tokens to qualified users and permanently records loan approval results. By providing verifiable audit trails, tamper resistance, and transparency, this blockchain layer increases confidence in decentralized financial (DeFi) services. Experimental findings confirm the hybrid architecture’s sustainability, scalability, and dependability. These findings focus on model accuracy and on-chain execution cost; broader claims about fairness and governance impact are discussed as architectural potential rather than measured outcomes. According to the study, AI-enhanced smart contracts can significantly enhance the security, accountability, and fairness of financial decision-making while providing a framework for future applications in supply chain management, healthcare, and decentralized identity management. Consistent with the scope described above, the fairness and governance benefits discussed here reflect the architecture’s design potential rather than experimentally demonstrated outcomes.