X is rewriting the source code for its search functionality, according to Feb. 21 post from Nikita Bier, head of product.
The move comes amid reports that X search has become less reliable in recent months.
Users have complained that X search is not returning relevant results and has become unusable in some contexts. Bier said in his post that his team is close to launching a new version of search, rebuilt to help handle the latest challenges.
As per Bier: “[Search has] been getting hammered by AI agents and has been choking at scale. We’re almost done with the full rewrite of legacy Twitter search (and also bot detection).”
X’s small engineering team (Bier recently said the team working on X’s day-to-day functionality is only 30 people) has been hard at work pumping out updates for the appm and the changes to search are just one of several coming changes that X hopes will help to address issues with spam and junk, as well as AI slop.
X also recently launched improved bot detection systems to weed out more junk, and it’s working on additional AI bot mitigation measures to get rid of the influx of automated AI replies in the app.
In a Feb. 20 post, Bier said that there’s no “silver bullet” to combat the rise of AI spam, and that the proliferation of AI tools will lead to problems for all social apps.
Which also impacts the value of the data being generated by users, which is then used to power AI models. For example, xAI uses X posts as its key data source to power Grok’s answers, but as that system starts ingesting more and more AI-generated posts, it eventually starts to degrade, and become a less valuable, less informative data stream.
Meta will be facing the same challenges, which is why all platforms are now working to eliminate overuse of AI. Yet, at the same time, these platforms are also prompting the use of their own AI tools to assist with post writing and content generation in order to maximize trend alignment and enhance overall activity.
In this sense, it’s a bit of a conundrum for Meta and X specifically. Both are investing hundreds of billions of dollars into their respective AI projects, so they need to promote their AI tools to get people excited about the potential (and assure investors that the outlay will be worth it).
But the overuse of AI will have negative impacts in terms of infecting models, annoying users, and overloading systems.
It’s another example of why AI and social media should not be intertwined, as AI tools run counter to the whole purpose of social media platforms as connective tools.
It’s too late to look back now, though, and with that in mind, the only option is for platforms to reform their systems in order to mitigate the negative impacts.
