Talk to the algorithm: Facebook is the latest platform to add a recommendation-altering chatbot for users

Major social platforms just love their chatbots–because those chatbots mean users are now responsible for fixing their own recommendation algorithms, instead of big tech companies having to (gasp) pay people for code cleanup.

YouTube did it, X did it, Instagram did it, and now Facebook joins the ranks. Its new “Your Algorithm” system asks users to tweak what they see on Facebook by telling an LLM topics they’re interested in, and by giving posts on their feeds a thumbs up or down.

“You can now fine-tune your experience and interact directly with the algorithms that rank and recommend what you see on Facebook, meaning you’ll see recommendations that are more relevant, more nuanced, and more you,” Facebook said in a company blog post.

An example shows Facebook’s chatbot telling a user, “Lately you’re all about good tunes, staying strong, salty ocean mornings, chasing great eats, and capturing every moment.”

The interests it recommends for them include music festivals, fitness, ocean kayaking, foodie adventures, videography, and grill recipes. Where the customization comes in is the little “+ Add” button at the end of the list for the user to tell it more topics they want to see in their feed.

There’s also a separate section for the user to tell it what they don’t want to see (in the example, banished topics are “hiking” and “horror movies”).

As for thumbs up and thumbs down on the feed, Facebook says tapping on the “Are you interested in this post?” button at the bottom of recommended content will bring up options asking the user to explain more about why they are or aren’t interested.

“You can easily tell us, in your own words, what you want to see more or less of, and your recommendations will update in real time,” the Meta-owned company says.

Facebook adds that it thinks its ranking system (aka its recommendation algorithm) was already “remarkably good” at “inferring what you’re into from what you engage with on Facebook.”

But “large language models are helping us take a genuine leap forward,” it believes.

“LLMs can incorporate the interests, context, and intent behind natural language, unlocking an algorithm you can actually talk to,” it says. “Under the hood, this marks a real evolution in how ranking works: by bringing freeform text together with everything we already understand about what you enjoy, the algorithms behind your Feed and Reels keep getting more relevant and responsive over time.”

In an era where constant exposure to drip-feeding algorithms can result in choice paralysis for users, it is interesting that platforms are now telling people to take back the reins by choosing what those algorithms ‘know’ about them.

This kind of system might also theoretically lessen the amount of personal information a platform would scrape from users (although that aspect will probably stay firmly in theory; we all know how salable private data is).

But, again, platforms might cite it in an attempt to free themselves from culpability for algorithm problems. If someone is regularly recommended harmful content, a platform with a chatbot could say, “Well, why didn’t you talk to the algorithm and disable that?” instead of altering the algorithm itself to prevent harmful content from even potentially reaching users.