Robo-advisors after the language model
Robo-advisors appear on 12 recent papers, and the topic is fading: 1 paper in July–September 2026 against about 5 per quarter before. Meanwhile people ask chatbots for financial advice. Does conversational advice change portfolios more than the old questionnaire-based robo-advice did, and for better or worse?
What prompted this#
Since October 2025, 16 papers carry a robo-advice keyword (8 in finance journals), with "robo-advisors" on 12. Only one of them was published in July–September 2026, against about 5 per quarter over the nine months before. "Household portfolio choice" appears on 15 papers. Generative AI, by contrast, is rising in almost every category on the trends page.
Questions#
- When households ask a chatbot about investing, do they get advice that fits their situation, and do they act on it more than on a robo-advisor's allocation?
- Does conversational advice reduce known mistakes (under-diversification, high fees, panic selling) or add new ones (overconfidence, chasing stories)?
- Should the same suitability rules that apply to advisers apply to chatbots?
Data that could help#
Experiments in which participants receive advice from a robo-advisor questionnaire, a chatbot or a human, with incentivized portfolio choices; brokerage data on users who adopt in-app AI assistants.
Next step#
Run an online experiment in which participants allocate a real bonus after advice from one of the three sources, and measure diversification and fees.