GENERATIVE AI AND INVESTMENT DECISION-MAKING AMONG INDIAN IT PROFESSIONALS: EMPIRICAL EVIDENCE USING MACHINE LEARNING-AUGMENTED STRUCTURAL EQUATION MODELLING

The proliferation of generative AI tools — including ChatGPT, Google Gemini, and Microsoft Copilot — in personal finance contexts marks a paradigmatic shift in how individual investors seek, process, and act upon financial information. This study investigates the impact of generative AI usage on investment decision quality among Indian IT sector employees through a two-wave longitudinal panel survey design (Wave 1: January 2025; Wave 2: July 2025; N = 520). Grounded in the Theory of Planned Behaviour (TPB) and Prospect Theory, the study employs a dual analytical strategy: Covariance-Based Structural Equation Modelling (CB-SEM) in IBM AMOS 26 for theory testing and machine learning classifiers (Random Forest, XGBoost, Logistic Regression) in Python 3.11 for predictive validation. Results from CB-SEM demonstrate that generative AI usage frequency significantly enhances financial self-efficacy (β = 0.443, p < .001), which in turn improves investment decision quality (β = 0.512, p < .001). Overconfidence bias significantly moderates the AI-decision quality relationship (β = -0.281, p < .01). Machine learning models achieve classification accuracy of 84.7% (XGBoost), outperforming logistic regression (72.4%), with SHAP analysis identifying financial self-efficacy and AI usage frequency as the two strongest predictors. The model explains 71.8% of variance in investment decision quality (R² = 0.718). These findings advance behavioral finance theory in the context of generative AI and offer actionable guidance for fintech developers, financial regulators, and corporate HR practitioners in India.