Artificial intelligence models used to detect depression on social media are often biased and methodologically flawed, according to a study led by Northeastern University computer science graduates.
Key biases in AI models used for detecting depression on social media
Reader’s Picks
-
The dynamics of family structure is undergoing a transformation. Declining fertility and mortality rates are leading to new kinship configurations [...]
-
When people think about family planning in Africa, they often picture women visiting clinics, women making decisions, and women bearing [...]
-
Emojis, as well as memes and other forms of short-form content, have become central to how we express ourselves and [...]