Microsoft
Research - [Research Intern, 2019]
My internship at Microsoft was my first experience in the industry so needless to say I was excited!
Misinformation and "fake news" were very prominent and concerning themes in society. Most past research on identifying misinfo hinged on black-box models. This could be good for model performance but difficult for clearly articulating the question of why? this is misinfo to an end user.
Our team applied existing concepts in deep learning and ML question-answering to build a novel approach to explaining why an article might be misinfo, empowering the user to be more informed and make the independent determination of whether they agreed with our reasoning.
In a time of increasing distrust of "Big Tech", the tech industry needs to further explore similar paradigms of "Hey, here's why we think this might be misleading." as compared to "Hey, we think this is misinfo. Don't ask how.".