4 Sources
4 Sources
[1]
Meta's 'Dear Algo' posts let Threads users tweak feed
Meta has decided to let Threads users make custom tweaks to its all-important algorithm, but don't expect your preferences to stick and do expect to bring your best manners. The social media-cum-metaverse-cum-AI firm formerly known as Facebook announced a new Threads feature called "Dear Algo" on Wednesday, alongside news of plans to build a massive 1 GW AI datacenter in Lebanon, Indiana. Dear Algo will use some of Meta's AI magic to tweak the algorithm for Threads users with just a public post, written like a letter directed to the algorithm, asking it to please show them more, or less of, a particular topic. Dear Algo posts can also be reposted by others, which will have the same effect on their account. Once published, the Dear Algo will tweak a user's feed per their request - but only for three days due to what Meta says is a desire to keep users "connected to the most current conversations" - or most controversial, whichever makes Meta more money. Meta didn't give many examples of the sort of commands users can make of Dear Algo, only mentioning things like telling it to show a person more posts about a current NBA game or something as general as "show me more posts about podcasts." The company didn't say how specific Dear Algo posts are able to get either, so whether a user could prompt it to exclude posts expressing certain viewpoints on a particular topic, or to mute spoilers about a particular television episode while allowing conversation about the show outside spoilers wasn't made clear. Meta's algorithms have been a source of controversy for years, ever since the company eliminated chronologically sorted Facebook feeds in favor of the company deciding how to present content to its users. Fast-forward more than a decade, and Meta is still taking heat for alleged racial bias in its algorithms, prioritizing content that may radicalize mass shooters, and generally promoting divisive, misery-inducing content. Personifying its algorithm by letting users write letters to it in a bid to help them get a more tailored experience for a few days may help appease some, but it's just as possible to leave others wondering what Meta could do to make its platforms more pleasant. Speaking of which, Meta didn't announce any plans to bring Dear Algo posts to its other platforms, so Instagram and Facebook users will have to continue to cope with an algorithm that provides what it thinks they want without an AI-powered option to change it.
[2]
Meta launches AI algorithm personalization feature for Threads
Meta on Wednesday debuted an AI feature called "Dear Algo" that lets Threads users personalize their content-recommendation algorithms. Threads users will be able to tell the Dear Algo tool what kinds of posts they want to see similar to how people use written prompts to interact with chat bots like OpenAI's ChatGPT. The feature is the latest example of how Meta is continuing to incorporate artificial intelligence into its various apps like Threads. The social media company on Tuesday also released AI features for its Facebook service that let users animate their profile photos and alter other images with the Meta AI digital assistant. Meta last month told investors that it plans to spend between $115 billion and $135 billion this year on AI-related capital expenditures. That's nearly double the amount of capex that Meta spent last year when it overhauled its AI unit. Dear Algo is the latest addition to Threads, the micro-blogging platform that Meta launched in July 2023 to compete against Twitter, which is now called X and was part of the merger between Elon Musk's xAI and SpaceX earlier this month. Last month, Meta said Threads had 400 million monthly active users and would begin rolling out ads globally. CEO Mark Zuckerberg in January also told analysts that Meta plans to debut and test new AI products and features in 2026 as the company continues to spend big on data center and related computing infrastructure. To use the new tool, Threads users have to to craft a public post on the platform that begins with the phrase "Dear Algo" and then explains the kind of content they want to see more or less of. A spokesman for the company said Meta was inspired by a trend of people publicly sharing posts that included the phrase "Dear Algo," a company spokesperson told CNBC. "Once you share your request, Dear Algo adjusts your feed for three days, so you can stay connected to the most current conversations,' Meta said in a blog post. "You can also repost someone else's Dear Algo request to apply their content preferences to your own feed." Meta said that it will begin testing Dear Algo with users in the U.S., U.K., Australia and New Zealand before rolling it out to more countries at later dates.
[3]
How 'Dear Algo' Feature Changes Meta's Threads Recommendation System
Meta has introduced a new AI-powered update for Threads called Dear Algo, which gives users a more direct way to shape what appears in their feed. Instead of relying only on automatic ranking, the feature allows people to publish a post beginning with "Dear Algo" and explain what they want to see more or less of. The platform's AI systems then interpret that request and adjust the user's feed accordingly for the next three days. At the same time, users can repost other Dear Algo prompts, effectively applying someone else's stated preferences to their own feed. As a result, feed tuning becomes temporary, shareable, and driven by explicit user input rather than long-term behavioural signals alone. For now, Meta has rolled out Dear Algo on Threads in the United States (US), the United Kingdom (UK), Australia, and New Zealand. However, the company says it plans to expand the feature to more countries soon. MediaNama tested the feature using a UK-based account. We published a post saying "dear algo give me more luxury content". The feature seemed to work initially with more UK-based English posts about luxury products; however, the algorithm started adding several posts in Tagalog, a language from the Phillipines, to the feed after the initial changes. Screenshots provided to MediaNama by a UK-based user showing posts in Tagalog, a language from the Phillipines. Source: MediaNama (MediaNama is trying to test feature with other UK-based users and we will upload more screenshots of their experience with the feature) During its Q4 FY25 earnings call, Meta said it plans to integrate large language models (LLMs) directly into the recommendation systems that power Facebook, Instagram, Threads and its advertising business. Chief Executive Officer (CFO) Mark Zuckerberg said today's recommendation systems are "primitive compared to what will be possible soon". Instead of apps that feel like "algorithms that recommend content", he said users will open Meta's platforms to find "an AI that understands you". According to Zuckerberg, the aim is to move beyond static engagement signals and build systems that interpret personal context, including a user's history, interests, content and relationships. He described this effort as building a "personal superintelligence" that can shape feeds, ads and commerce around individual goals. In effect, Meta is positioning LLMs as the core infrastructure for ranking, discovery and ad delivery across its services. Beyond recommendations, Meta said it is deploying agentic AI across commerce and business tools. The company is developing shopping agents that surface specific products from its catalogue across feeds and messaging, including WhatsApp. It is also testing a Meta AI assistant for advertisers to support campaign optimisation, while expanding business AI chats in markets such as Mexico and the Philippines. Meta has now expanded advertising on Threads to users in all markets worldwide, integrating the platform fully into its advertising infrastructure and allowing brands to run campaigns through Meta's Ads Manager alongside Facebook and Instagram. The ads, which support both image and video formats and use Meta's existing buying tools, including Advantage+ placements, will appear directly in users' feeds as part of a gradual global rollout. This expansion comes as Threads continues to scale its audience. Since launching in July 2023, the platform has crossed 400 million monthly active users (MAUs), according to Meta, marking substantial growth from earlier earnings calls where CEO Mark Zuckerberg reported figures of around 150 million MAUs in early 2024 and 275 million later that year. Despite this progress, Threads' daily active user count remains a small share of Meta's overall footprint, which spans billions across Facebook, Instagram and WhatsApp. With advertising now rolled out globally on Threads and Dear Algo inviting users to explicitly state what they want to see, an obvious question follows: will Meta now gain even richer insight into user preferences? On the one hand, Dear Algo appears to give users more control over their feeds. However, it also generates structured signals about interests, dislikes and intent, signals that Meta's systems can interpret at scale. At the same time, Meta has made clear that it plans to integrate LLMs directly into its recommendation and advertising infrastructure. Rather than relying solely on passive engagement metrics, the company wants AI systems that understand personal context, relationships and goals. Consequently, combining explicit preference prompts with global ad delivery could strengthen Meta's ability to refine targeting, ranking and commerce. In effect, Threads may now function not just as a social feed, but as a live testing ground for Meta's broader ambition: AI-driven systems that shape discovery, advertising and shopping around increasingly granular user data.
[4]
Meta introduces "Dear Algo", AI to personalize the Threads feed
Meta has unveiled a new artificial intelligence feature called "Dear Algo," designed to give Threads users greater control over their recommendation feed. With the tool, platform members can set content preferences as written prompts, similar to those addressed to chatbots, to temporarily adjust their user experience. To activate it, users simply publish a public post starting with "Dear Algo," followed by their desired preferences. The algorithm then adapts the news feed for three days. Users can also reuse a request posted by another user to apply the same criteria to their own feed. The feature is directly inspired by an already common practice on the platform, where users tried to influence the algorithm with messages addressed to "Dear Algo." The rollout fits into Meta's broader push to integrate AI across its services. In parallel, the company announced new Meta AI-based features, such as animating profile photos or editing images on Facebook. Meta plans to invest between $115bn and $135bn in AI in 2026, a sharp increase from the previous year. Threads, launched in July 2023 to compete with X (formerly Twitter), had 400 million monthly active users in January. The platform is also beginning a global advertising rollout. "Dear Algo" will first be tested in the United States, the United Kingdom, Australia and New Zealand, before being gradually expanded to other markets.
Share
Share
Copy Link
Meta introduced Dear Algo, an AI-powered feature that lets Threads users personalize the Threads feed through written prompts. Users can post requests starting with "Dear Algo" to influence content recommendations for three days. The feature is now testing in the US, UK, Australia, and New Zealand as Meta pushes deeper into AI algorithm personalization.
Meta has launched Dear Algo, an AI-powered feature that gives Threads users a new way to personalize the Threads feed through direct communication with the platform's content-recommendation algorithms
1
2
. The tool allows users to craft public posts beginning with the phrase "Dear Algo" followed by written prompts that specify what content they want to see more or less of, similar to how people interact with chatbots like OpenAI's ChatGPT2
. Once published, the Dear Algo request adjusts the user's feed for three days, with Meta citing a desire to keep users "connected to the most current conversations"1
.
Source: The Register
A Meta spokesperson revealed the company was inspired by an existing trend of users publicly sharing posts with the phrase "Dear Algo" in attempts to influence content recommendations
2
. The feature also enables users to repost someone else's Dear Algo request, effectively applying their content preferences to their own feed2
3
. This makes feed tuning temporary, shareable, and driven by explicit user input rather than long-term engagement signals alone3
.Meta has rolled out Dear Algo in the United States, United Kingdom, Australia, and New Zealand, with plans for a global rollout to additional markets
2
3
. However, early testing by MediaNama using a UK-based account showed inconsistent results in the Threads recommendation system3
. When a user posted "dear algo give me more luxury content," the feature initially delivered more UK-based English posts about luxury products, but the algorithm subsequently began adding several posts in Tagalog, a language from the Philippines3
.
Source: MediaNama
Meta provided limited examples of how specific Dear Algo commands can be, mentioning only general requests like showing more posts about a current NBA game or podcasts
1
. The company did not clarify whether users could prompt the system to exclude posts expressing certain viewpoints or mute spoilers about specific television episodes1
. This AI algorithm personalization feature remains exclusive to Threads, with no announced plans to bring it to Instagram or Facebook1
.The Dear Algo launch coincides with Meta's ambitious AI strategy outlined during its Q4 FY25 earnings call. Mark Zuckerberg stated that the company plans to integrate Large Language Models (LLMs) directly into the recommendation systems powering Facebook, Instagram, Threads and its advertising systems
3
. Zuckerberg described today's recommendation systems as "primitive compared to what will be possible soon," envisioning platforms where users encounter "an AI that understands you" rather than simple algorithms3
.Source: Market Screener
This vision involves building what Zuckerberg calls "personal superintelligence" that interprets personal context, including user history, interests, content and relationships to shape feeds, ads and commerce around individual goals
3
. Meta plans to spend between $115 billion and $135 billion in 2026 on AI-related capital expenditures, nearly double last year's spending when the company overhauled its AI unit2
. The company also announced plans to build a massive 1 GW AI datacenter in Lebanon, Indiana1
.Related Stories
The introduction of Dear Algo raises questions about how Meta will leverage the structured user preference data generated through these explicit written prompts
3
. While the feature appears to give users more control over their user experience, it simultaneously generates rich signals about interests, dislikes and intent that Meta's systems can interpret at scale3
. This development comes as Meta has expanded advertising on Threads to users in all markets worldwide, with the platform now serving as a testing ground for the company's broader AI ambitions3
.Threads, launched in July 2023 to compete against X (formerly Twitter), reached 400 million monthly active users in January and began its global rollout of advertising
2
. Meta's algorithms have faced controversy for years over alleged racial bias, prioritizing divisive content, and promoting material that may radicalize users1
. The ability to tweak feed preferences through Dear Algo may help address some user concerns, though the temporary three-day window and unclear specificity limits leave questions about the feature's long-term impact on content moderation and user satisfaction.Summarized by
Navi
[1]
[4]
23 Nov 2024•Technology

28 Sept 2024

29 Jan 2026•Business and Economy

1
Policy and Regulation

2
Technology

3
Technology
