2024 Global Retail Investor Outlook 2025

AI can revolutionize financial services by enhancing accessibility, but its bias and accuracy risks must be addressed Large language models (LLMs) are trained on historical financial data that may lack sufficient representation of underserved populations. This can lead to biases in AI systems, limiting access to financial services for marginalized groups.47 To mitigate AI bias, it is key to ensure rigorous data hygiene, diverse training datasets and due diligence. Given AI’s tendency to “hallucinate” and generate inaccuracies, especially in financial advisory, awareness of these limitations is crucial.48 Integrating AI into financial processes faces challenges in scalability, data governance and security As AI matures, scalability challenges demand robust infrastructure to support a growing user base.Data governance is crucial for ensuring database hygiene, and so is establishing clear use and protection guidelines. Safeguarding proprietary financial data prevents breaches and protects sensitive information.49 In total, 49% of investor respondents find that barring the ability to access a personal wealth adviser, advanced security features and data protection are key to establishing trust in a financial institution. There are concerns about unintentional personal data disclosures by both retail investors using AI tools and financial institutions managing sensitive information. The ability to trace and reproduce AI- generated results remains a critical technical challenge as tools are embedded into processes like risk management and market operations.50Incorporating AI solutions to accelerate capital markets’ democratization comes with inherent complexitiesCASE STUDY 10 Mitigating AI bias and shortcomings The European and Securities Market Authority (ESMA) has the goal of systematically ensuring transparency, responsible governance and individual investor protection, requiring financial institutions to disclose AI use in investment decisions per MiFID II. ESMA stresses the importance of accurate data to train AI models, and advocates for continuous financial staff training on AI’s operational and regulatory implications. It also requires stress testing and quality assurance checks for AI tools.51 By doing this, ESMA plans to mitigate AI-linked risks, such as algorithm bias, sensitive data breaches and lack of outcomes explainability.CASE STUDY 11 AI to support open banking and lessen system fragmentation Launched by the Reserve Bank of India in 2016, the Account Aggregator (AA) network streamlines data sharing among financial institutions to reduce fragmentation and expand banking access, including lending and investments. While AA only facilitates encrypted data transfers, its design supports key AI applications, from natural language processing (NLP)- and optical character recognition (OCR)- driven data extraction to ML-powered credit scoring and fraud detection, driving India’s open finance ecosystem. 42% 18%of current investors would invest more if they had an AI chatbot assisting them.of non-investors would consider investing more if they had an AI chatbot assisting them. 2024 Global Retail Investor Outlook 58