My dissertation research looked at the ways that the design of intelligent decision aids (decision aids that use artificial intelligence or similarly sophisticated computation to make recommendations to decision makers) can create decision making biases.
I found that the customizability of the system, the transparency, and users expectations of its efficacy can cause users to agree with system recommendations regardless of what recommendations are given.
Going forward, I plan to extend this research to the design of tools for data science. How can we design tools that enable data scientists to both uncover important insights and make good decisions from data?
Heterogeneity in Customization of Recommender Systems By Users With Homogenous Preferences (CHI '16)
Decision Biases in Agreement with Intelligent Decision Aids (Dissertation)