Machine learning models are useful and fair only when utilized correctly by human decision-makers. In an age where data is usually biased and machine learning tools are not perfect, people need to know when and when not to rely on automated machine learning results for real-life decisions. We aim to design and implement an exemplar-based teaching strategy to teach people when to rely on the ML model’s decision versus their own by picking up on patterns with ML performance (accuracy). We will develop an algorithm to provide feedback on the person’s performance in assessing the accuracy of the AI during training, making the teaching process interactive.

I really enjoyed the machine learning courses I took in the past and would like to further explore the field of machine learning and human interaction research. I am looking forward to applying my ML and statistics knowledge to this project and producing a paper that is interesting and impactful. I am excited to become a better researcher with strong technical and soft skills by participating in SuperUROP.