Platforms such as Instagram, Facebook, TikTok, and YouTube are expanding their tools to enable the personalization of suggestions presented in their feeds. This change comes in response to criticism from users who feel limited by automatic recommendations that do not always match their interests.
Thanks to advances in artificial intelligence (AI), these platforms are now capable of processing more detailed requests, offering users new ways to direct what is displayed on their screens.
Historically, platforms prioritized automatic recommendations over posts from accounts that users followed. The initial goal was to anticipate content that would spark interest and keep users connected longer. However, this model was also criticized for directing users to sequences of irrelevant or potentially harmful videos.
Instagram is seeking to modify this dynamic through a feature called 'your algorithm.' This functionality displays topics that the network associates with each individual's interests and allows the user to indicate themes they would like to see with higher or lower frequency. Initially available in the Reels tab, the function has been extended to the entire application and also incorporated into Facebook.
Tessa Lyons, Vice President of Product at Instagram, told CNN that the goal is to restore some of the control users had in selecting which profiles to follow, without abandoning useful recommendations for discovering new content.
Social networks have adopted different approaches to meet this demand. Neal Mohan, CEO of YouTube, emphasized that the changes do not imply the impossibility of viewing any AI-generated content, arguing that differentiating between videos created by artificial intelligence and those produced by humans is an artificial distinction, given that many productions blend both forms.
More direct personalization depends on technology capable of understanding natural language commands. Instead of just inferring all preferences based on views and interactions, systems can interpret specific requests about which subjects should appear with higher or lower occurrence.
Lyons commented to CNN that this advancement allows preferences to be expressed using one's own words. She explained that previously, platforms struggled to accurately identify which posts related to specific topics, such as tennis.
However, a challenge remains: online behavior does not always reflect what people truly want to see. Tim Weninger, a professor of computer science and engineering at Notre Dame University, points out that content that evokes feelings of anger or sadness can also generate engagement, such as views, comments, and shares.
These interactions can confuse recommendation mechanisms, which need to discern between genuine interest and mere emotional reaction. The big question is determining to what extent users' stated preferences will be able to influence algorithms compared to the signals captured during browsing.