AI & Financial Advice: Will LLMs deliver trusted financial advice?

TLDR

Research-based insights exploring how Large Language Models (LLMs) are reshaping financial advice - from democratizing access to enhancing personalization while helping to manage new risks and responsibilities as the investor’s new AI fiduciary

In brief

Recent research indicates that Large Language Models (LLMs) are reshaping the financial advice landscape with promising capabilities and adoption trends. Investors are already embracing AI tools for financial guidance, with 29% currently using tools like ChatGPT to access financial information and recommendations according to the Ontario Securities Commission (2024).

The research highlights five essential pillars emerging in what Tagor AI calls "Advice Intelligence":

1. Human-AI partnership: Studies show investors follow blended human-AI advice more closely than human-only guidance, with 9% higher adherence to investment recommendations in controlled experiments by the Ontario Securities Commission. While AI reduces gender bias in recommendations, it still exhibits home bias, emphasizing the need for human oversight.

   
   

2. AI systems over models: The future requires complete systems, not just powerful models. MIT research (2024) confirms that supplemental modules combined with LLMs have the acumen to provide financial advice. Effective systems coordinate specialized components with LLMs serving as the intelligence layer rather than standalone advice engines.

   
   

3. Trust engineering: Success hinges on building reliable, transparent AI systems that explain their decisions. Charles Schwab (2024) found that 75% of 21- to 26-year-olds express comfort with AI-assisted financial planning. Clear explanations of data sources, methods, and limitations significantly increase investor confidence.

   
   

4. Adaptive intelligence: Larger foundation models outperform specialized smaller models in financial advisory tasks. A 2023 study by the Munich Society for the Promotion of Economic Research found that GPT-4 portfolios generally provide exposure to the same geographies and asset classes as professionally advised portfolios

   
   

5. Democratizing high-quality advice: LLM-powered systems can provide personalized advice at significantly lower costs while maintaining consistent quality across different client segments, potentially making sophisticated investment strategies accessible with reduced minimum investment requirements.

   
   

Looking ahead, researchers, regulators, and practitioners must collaborate to develop frameworks for AI-driven financial advice that keeps pace with adoption. Key areas for future work include multi-modal interfaces, model ensemble approaches, AI-investor collaboration models, and autonomous AI agents that balance accessibility benefits with potential risks.

While progress continues, these systems still require refinement - they're showing meaningful advancement toward democratizing financial advice, but the journey from progress to perfection is ongoing.

AI&FinancialAdvice_TagorAI_Published.pdf

Download PDF • 647KB