Systematic review uncovers pathways to extremist and radicalizing material in YouTube recommender algorithms, suggesting algorithmic exposure risks despite modeling limitations.
Key Points
To determine whether YouTube's recommender system facilitates user pathways to problematic content, such as extremist or radicalizing material.
Conducted a systematic review and narrative synthesis adhering to the PRISMA guidelines.
Screened 1,187 initial studies to yield a final analytical sample of 23 eligible papers.
Of the 23 evaluated studies, 14 implicated the YouTube recommender system in facilitating problematic content pathways, seven reported mixed results, and two found no implication.
Evidence indicates recommendation algorithms can direct users toward problematic content, though researcher models are constrained by limited visibility into YouTube's proprietary algorithms.