Today, recommendation engines are the biggest threat to societal cohesion on the internet—and, as a result, one of the biggest threats to societal cohesion in the offline world, too. Recommendation engines have become The Great Polarizer. Today’s world and countries are polarized to much extent, Much of it can be credited towards misinformation campaigns that use recommender systems to amplify and maximize the fringe thoughts and feelings one may have.
One simple way is just to add more randomization to the model which would help mitigate spread of misinformation, but that would take away the accuracy and profitability of a particular model, and it would also take away from the important positive user experience that the model also employs.
Social influence bias can be significant in recommender systems and that this bias can be substantially reduced with machine learning. One challenge is to apply this to large datasets. One way is to cluster/classify items into a small number of representative categories and train a model for each category.
Another way to potentially improve upon these systems is to add more verification information to them. Maybe a video/audio can be scanned and with natural language processing and checked for facts and if the viode/audio is found to be compromising the facts the video could be ignored by the recommender system and not presented to the user interacting with the recommender system.
Ethics needs to become more prominent in recommender systems rather than focus only on profitability and the responsibility lies with all companies uilding and implementing the recommender systems and moreover as user one needs to be more informed about how to use the recommender systems work and shold be given options on whether the user wants to include the curiosity clicks versus genuine clicks. Althought this mght be more complex to implement i hope in future this methodology would be more common.