Robo-advisors, which automate financial guidance, put Wall Street expertise within reach of Main Street savers. But should these sophisticated algorithms be left to manage portfolios without us? Is there value in keeping humans in the loop? In a five-year study tracking 38,000 French investors, TSE researcher Milo Bianchi and coauthor Marie Brière find surprising answers.
What motivated you to study how humans interact with robo-advisors?
Robo-advisors promise to democratize wealth management, bringing professional-grade portfolio advice to ordinary savers at a fraction of traditional costs. But there’s a catch… When algorithms take full control, do people learn anything? Do they stay engaged? Or do they just hand over the keys and hope for the best?
We still know relatively little about how people respond to algorithmic guidance. For instance, what’s the optimal degree of automation? Existing research suggests people are more willing to rely on automated advice when they retain some control. Human-robot interactions can also be useful for promoting investors’ learning on how to manage their portfolios. At the same time, letting the human interfere with the robot may limit its effectiveness.
Understanding these trade-offs is essential for responsible regulation and design of AI-based financial tools. Given the widespread use of fully automated systems, we wanted to explore the impact of robo-advisors when investors retain the final say.
How does your paper investigate this topic?
Following the introduction of a robo-advisor by a large French asset manager, we study the impact on real portfolios from 2016 to 2021. Our dataset covers 770 firms, including more than 18,000 subscribers and 20,000 “robo-curious” investors who showed an interest in the service but declined it. This unusually detailed setting lets us observe how actual investors behave—with their own money and long-term saving plans—when given algorithmic advice.
Unlike fully automated services, the robot issues portfolio recommendations and rebalancing alerts while leaving investors free to accept, ignore, or adapt them. By tracking these human-robot interactions over time, we can determine whether algorithms improve or crowd out human attention and judgment.
Do investors who use robots take their eye off the ball?
Our evidence points the other way. The robo-service sparks more investor attention, not less. After taking up the service, investors log in more often, spend more time on the platform, make more portfolio adjustments in response to market shocks or new opportunities, and increase their exposure to risk.
Following alerts, service users increase their rebalancing activity to keep their investments more closely aligned with their long-term goals. These behavioral changes translate into significantly higher returns.
Why do ‘human-robot interactions’ play such a vital role?
We show that the process of engaging with a robo-service is at least as important as the content of the robot’s recommendations. Robo-alerts act as attention triggers, prompting investors to reconsider allocations, and potentially to learn. Think of it like a fitness tracker that counts your steps—its real value lies in making you notice your activity patterns and adjust your behavior. Similarly, robo-alerts prompt investors to reflect on whether current allocations still match their long-term goals.
Surprisingly, when we simulate what would have happened if the robot had complete control—automatically rebalancing without human input—the performance gains are only marginally better. In other words, keeping humans in the loop costs very little in performance terms while offering important potential benefits for financial learning. This suggests we should not leave all decisions to the robot, instead allowing dynamic human-robot interaction to serve as an educational tool.
For AI doomers, your results may be reassuring. Are there important caveats?
By nudging investors to engage more and learn about rebalancing, robo-advisors may help turn passive savers into more active and capable portfolio managers. However, our study tracks behaviors over five years, not decades. Whether this investor engagement translates into lasting financial sophistication remains an open question.
More concerningly, the system works less well during market downturns. When share prices are falling, investors become significantly less likely to follow rebalancing alerts. This suggests the robot's educational benefits may be fragile in bear markets, when disciplined rebalancing matters most.
Another important aspect is that effects can vary significantly across investors. For instance, older, male, and wealthier investors are slightly more likely to rebalance after a robo-alert.
What are the lessons for service providers, employers, and regulators?
For providers and employers, the design of human-centered robo-services is key. Alerts, clear recommendations, and retention of human choice can yield meaningful gains. For regulators and consumer protection agencies, the message is that automation works best as a partnership rather than an all-or-nothing substitution. Robo-advisors should not replace human judgment; they should educate and empower it.
KEY TAKEAWAYS
• Robots wake people up – Subscription to the robo-service increased logins, platform engagement, and trading activity.
• Nudges work – Robo-alerts raise the probability of rebalancing. When investors rebalance, they typically follow the robot.
• Performance gains – Robo-users enjoy higher returns than non-users. Fully automated rebalancing only slightly outperforms human-robot rebalancing.
• Hybrid future – The sweet spot combines algorithmic precision with human oversight, enhancing both financial returns and financial skills.
FURTHER READING TSE Sustainable Finance Center promotes innovative research on emerging issues in economics and finance. Human-Robot Interactions in Investment Decisions and other publications by Milo Bianchi and Marie Brière are available on the TSE website.
Article published in TSE Reflect, November 2025
