However, removing human writers from Trending doesn’t necessarily eliminate bias. Human bias can be embedded into algorithms, and extremely difficult to strip out. That’s one of the conclusions from a study (pdf) of a popular algorithm used for processing language from Princeton University and the University of Bath released as a draft yesterday (Aug. 25). It’s currently under review for publication in a journal.
“Language itself contains recoverable and accurate imprints of our historic biases,” the authors write. “Whether these are morally neutral as towards insects or flowers, problematic as towards race or gender, or even simply veridical, reflecting the status quo for the distribution of gender with respect to careers or first names.”
Facebook’s increased reliance on automation reflects the company’s growing faith in machine intelligence. But even machines may fail to overcome some the most fundamental problems that humans face, like operating fairly in the face of embedded bias.