A systematic review: The YouTube recommender system and pathways to problematic content
By Muhsin Yesilada, Stephan Lewandowsky. 2021
9 citations recorded by the source corpus.
The systematic review aimed to determine whether the YouTube recommender system facilitates pathways to problematic content such as extremist or radicalising material. The review assessed the eligibility of 1187 studies and excluded studies using the PRISMA process for systematic reviews, leaving a final sample of 23 studies. Overall, 14 studies implicated the YouTube recommender system in facilitating problematic content pathways, 7 produced mixed results, and two did not implicate the recommender system in facilitating pathways to problematic content. Thus, the review's findings indicate that the YouTube recommender system could lead users to problematic content via the recommender system. The review indicates a need for corrections to the recommender system to shift users away from problematic content.
Subfields: Artificial Intelligence, Information Systems, Sociology and Political Science, Recommender Systems, AI Ethics, Fairness & Society
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Iason Gabriel
- Po-Sen Huang
- Jonathan Uesato
- William S. Isaac
- John F. J. Mellor
- Borja Balle
- Geoffrey Irving
- Myra Cheng
In Information Systems
- William S. Isaac
- Iason Gabriel
- Borja Balle
- Myra Cheng
- Po-Sen Huang
- Atoosa Kasirzadeh
- Abeba Birhane
- Jonathan Uesato
In Sociology and Political Science
- William S. Isaac
- Iason Gabriel
- Borja Balle
- Myra Cheng
- Lisa Anne Hendricks
- T. Stepleton
- Atoosa Kasirzadeh
- Abeba Birhane
In AI Ethics, Fairness & Society
Citations in the corpus
- Ethical and social risks of harm from Language Models (2021, background): “…switch to a human once the victim 11Some recommender systems have been found to respond to certain user behaviour by recommending more and more extreme viewpoints to increase engagement ((O’Callaghan et al., 2015; Yesilada and Lewandowsky, 2021); for counterexamples view (Möller et al., 2018)).”