It is certainly true that recommender systems can have a tendency to funnel a user into a feedback bubble which can have an overall detrimental effect the sharpest example of this is in the New York Times article. It does not help that we encounter such recommendations everywhere on any technology platform. However, tube sites and social media platform seem to be the most dangerous in this category as the bubble that the user can get engulfed in can be “radical” and lead to a skewed version of reality.
The root of the problem is that these platforms are a business, and a business needs to have a steady stream of customers. The recommendation algorithms are optimized to do just that. The obvious answer to this is content moderation but this is a deeply loaded political, ethical, and subjective question.
One way to perhaps combat this situation is providing diversity in the recommendations. This keeps the element of novelty as well as providing the user with alternative points of views or other items.