Princeton and Boston University researchers have found evidence that the emerging field of artificial intelligence may not be immune to implicit bias. Inspired by the implicit association test (IAT), researchers used algorithms to identify associations based on hundreds of billions of words on the internet and found similar results to the IAT. They also found that Google's translation software converts gender-neutral pronouns from several languages into “he” for doctor, and “she” for nurse.
A year-long study demonstrated the efficacy of women engineering students being mentored by another woman engineer. Nilanjana Dasgupta, from the University of Massachusetts in Amherst, found that the mentored women were motivated, self-assured and less anxious than those paired with male mentors or having no mentor at all. She describes mentors as “social vaccines,” preserving the students’ confidence and feeling of belonging. Of the women followed during the study, all of those with women mentors stayed with the program while their non-mentored counterparts saw an 11% dropout rate.