Twitter Implements Trend Deletion and AI News Summarization

Twitter, now X, is currently experimenting with a new section called “Explore”, instead of TT (Trending Topics). This is a summary of four news stories by Grok, Elon Musk's generative artificial intelligence. No one will see the same trends anymore.

When we click on one of these news items (personalized according to the users' algorithm), we do not see tweets, but a paragraph. This is a summary of the most popular posts, by Grok artificial intelligence, developed by xAI. Below, the tweets that were used to generate this summary appear.

Screenshot of Twitter's new Explore menu, with an article summarized by Grok on the right. // Source: Numerama Captures

Information told by an AI: what could go wrong?

The Explore feed is never the same. Here with four journalists from Numerama. — The Explore feed is never the same. Here the sons of four journalists from Numerama. // Source: Numerama Captures

In addition to these more exhaustive titles, X offers an AI summary when you click on a subject. Enough to cause concern, when popular tweets on Twitter are often divisive.

Is there a chance that Grok is confusing opinion with information? The first articles summarized by the social network often concern Elon Musk, Tesla and SpaceX. An article on Russia leans in favor of the country, which would be supported by users of the social network for its posture towards Israel… Some tweets put forward by Grok come from far-right accounts.

Based on tweets, Grok sees here that users support Russia for its opposition to Israel. // Source: Capture Numerama

“This news is a summary of messages posted on X and may change over time. Grok can make mistakes, check his statements »indicates each news summary, to clear X of any responsibility in the event of fake news.

If the Explore function is available in France, please note that the titles and articles summarized are only in English. Twitter does not offer a translation tool, which suggests that its deployment in France may be an error. Grok, in its chatbot form, is not available from us.

A test likely to become widespread?

Can replacing trends with summaries of popular tweets be sustainable? There is a risk for X, often singled out for its policies which accentuate disinformation. Paid accounts are better valued, which allows them to express their opinions more easily. If Grok is unable to question the origin of information, it could turn into a vast disinformation tool, telling nonsense to users of the social network, and what's more, targeting them with content they want to hear.

For the moment, Elon Musk and X have not communicated about this test, which could take time to definitively eclipse the trends. The billionaire nevertheless considers that X will replace the media, which suggests that this test would be part of a broader vision.