The link above is a report I wrote with Caterina Giannetti about some research we did that was financed by the Think Forward Initiative (https://www.thinkforwardinitiative.com).
We were investigating the use of robo-advice, that is, advice that is given by algorithms rather than by humans. This has great potential to help people make better decisions cheaply. However, in the special case of stock trading, we found that people were decidely not very enthusiastic (only about 30% wanted to use them).
However, we also found that people preferred active vs. passive robo-traders, and they liked to be able to override decisions made by the algorithm. There was also a typical Dunning-Kruger effect, as those who would have profited from advice the most were the least willing to get it.
The experiment was nicely designed so people traded via their smartphone over three weeks, with 3 trading periods per day, and 3 stocks to trade each period. Participants thereby gained lots of experience. We had high participation (trading 2 times per day on average) and low drop-out rate (6%).
The analysis we report at the link above is still very preliminary. In the coming academic paper, we will focus on who gains from using robo-advisers and who avoids them. We will go deeper into individual differences, and how adoption of a robo-adviser depends on trading style and events during preceding periods.