Instagram has a means of showing you stuff that other people “did for the ‘gram” that uses AI to tailor content to your tastes.
The social media app shared the details on this, saying that it tries to populate the ‘Explore’ tab with recommendations for whole accounts, not just individual posts.
Algorithmic recommendation systems and machine learning in general have received scrutiny for habits like showing users extremist content and containing biases.
More from a recent post on The Verge:
While Instagram has not been criticized with the same ferocity as YouTube (dubbed “the Great Radicalizer” by The New York Times), it certainly has its share of problems. Hateful content and misinformation thrive on the platform as much as any other social network, and certain mechanisms in the app (like its suggested follows feature) have been shown to push users toward extreme viewpoints for topics like anti-vaccination.
Instagram AI faces a key challenge in recommending content to its users: chiefly, the fact that the platform’s content ranges so widely.
Related: Instagram is Taking Phishing Attacks Seriously
The solution, the platform says, is to focus on accounts which might interest users. They do this by identifying accounts which are similar by adapting “word embedding” machine learning to study the order in which words appear in text.
The Verge gives this example:
A word embedding system would note that the word “fire” often appears next to the words “alarm” and “truck,” but less frequently next to the words “pelican” or “sandwich.” Instagram uses a similar process to determine how related any two accounts are to one another.
The Explore system begins by looking at “seed accounts,” which are accounts that users have interacted with in the past by liking or saving their content. It identifies accounts similar to these, and from them, it selects 500 pieces of content. These candidates are filtered to remove spam, misinformation, and “likely policy-violating content,” and the remaining posts are ranked based on how likely a user is to interact with each one. Finally, the top 25 posts are sent to the first page of the user’s Explore tab.
The only problem with Instagram’s technique is that they didn’t reveal what signals are used to identify spam, though obviously sharing that information could ultimately help spammers.
