Financial Education in the Age of Artificial Intelligence: A Systematic Review with Text Mining and Natural Language Processing

This article develops a rigorous and reproducible systematic review of the integration of artificial intelligence (AI) in financial education during the period 2020–2025, structured in accordance with -5.3-PRISMA and explicitly oriented toward detecting narrative and perception. The search was conducted in three complementary databases (Scopus, ScienceDirect, and Taylor & Francis), using search strings equivalent to those of the platform and a selection workflow that begins with 388 records and culminates in 50 included studies, prompting a narrative synthesis given the methodological heterogeneity. From a methodological contribution perspective, the study combines bibliometric mapping with text mining and an NLP process that triangulates sentiment using lexicon-based approaches (VADER, TextBlob) and a multilingual transformer model (XLM-RoBERTa), producing continuous indicators (sentiment index) and reproducible research artifacts. The results position AI as an integrative nexus linking financial literacy, decision-making, sustainability, and language technologies (including ChatGPT-5.3.), highlighting its potential for personalization, virtual tutoring, and immediate gains in comprehension and motivation; however, evidence of sustained behavioral change remains nascent. Critical gaps remain, such as a shortage of longitudinal/controlled studies, a lack of standardized metrics, limited transparency and validation of models, and constraints in terms of geographic and cultural diversity, while privacy, fairness, and algorithmic bias emerge as structural conditions for responsible adoption.