Instagram explains how its algorithm suggests new content

After Meta CEO Mark Zuckerberg announced plans to more than double the amount of “recommended” content that appears in feeds by the end of 2023, Instagram has revealed how its algorithm suggests new content to users. .

On Friday, Instagram’s engineering team released a detailed look at how the app’s artificial intelligence (AI) selects suggested posts that users see in their feed.

Instagram launched suggested posts in August 2020. According to the team, the goal was to replicate human discovery of new content on Instagram with AI features, to improve user engagement.

Instagram’s engineering team says the app’s posting recommendations fall into two categories: “Connected” and “Not Connected.”

In the connected system, Instagram’s AI finds and highlights recommended posts based on the user’s interests. The ranking system then selects the best posts based on factors such as engagement, relevance, content quality, and freshness.

In an unconnected system, sources are derived implicitly based on a user’s activity on Instagram and are then ranked based on similar factors.

The process relies heavily on implicit signals from actions users have taken directly in the app, such as following and liking posts. But it can also extend to who users follow and what they like, as a proxy for direct engagement, while some popular posts are also highlighted based on overall engagement.

According to the engineering team, the intention is to replicate the feel of the posts and profiles that Instagram users have chosen to follow when viewing suggested posts – to make them feel familiar.

“Scrolling through end-of-feed recommendations should feel like scrolling through an extension of the Instagram home feed,” the team says.

The team says that if users don’t have enough recent engagement on Instagram for the AI ​​to suggest posts for them, the system will evaluate accounts followed by that user and use them as seeds for suggested content.

If the user is new to Instagram, the app simply shows them popular media and adapts the settings based on their first responses.

Instagram’s engineering team says it tries to ensure that the mix of media types, such as video and traditional posts, for example, is relatively similar in relation to the homepage. and suggested publications. However, this post appears to be at odds with Instagram’s recent move to push more videos, especially Reels, into user feeds.

Instagram has faced a lot of criticism from its users for its recent changes. The growing backlash forced Instagram manager Adam Mosseri to reverse the changes and temporarily reduce the volume of recommended posts on the app so the team could improve the algorithm.

But Instagram users can expect to see the number of suggested posts steadily increase over the next year. During Meta’s second quarter earnings call last month, Zuckerberg revealed that currently around 15% of the content presented to a Facebook user on his feed is recommended by his algorithm – the number being slightly higher. on the Instagram feed. However, according to Zuckerberg, this number is expected to increase to more than 30% by the end of 2023.

“Social feeds are no longer primarily driven by the people and accounts you follow, but are increasingly driven by AI recommending content you find interesting on Facebook and Instagram, even if you don’t follow those creators. “says Zuckerberg. .

Comments are closed.