An approach to detecting fake and misleading content on YouTube for the older generation in Bangladesh, using HCI and multilingual NLP models

Abstract

The rapid advancement of digital technologies has dramatically impacted how people access information and connect globally. However, this revolution presents specific challenges, especially for the older generation, who are more prone to fake and misleading content due to their lack of proper digital literacy and awareness. Easier access to the internet and a user-friendly interface for YouTube make it a popular choice amongst the older generation of social media users. However, YouTube’s user-friendliness and weak content validation process make it easier for malicious users to deceive viewers and capitalize on misinformation. In this study, we explore the impact of such manipulation on older social media users in Bangladesh, focusing on YouTube. By leveraging Human Computer Interactions (HCI) methodologies, we analyze user behavior to discover how YouTube’s deceptive content influences older users’ behaviours. Also, we identify vulnerabilities and propose strategies to increase digital safety for this age group. In this research, we conduct surveys and interviews to identify the types of fake and misleading content that older users encounter and evaluate their ability to identify and protect themselves against it. To address these challenges, we have experimented with NLP models to create a browser extension-based system to enhance digital safety for this age group. Our research aims to improve the YouTube Interface to be more user-friendly while implementing robust and effective content validation processes, ensuring everyone, especially older generations, can navigate YouTube confidently and securely.