I’ve been trying for quite some time to reduce my time spent on X. Nevertheless, in a bid to not focus solely on the dismal state of everything, I will mention something positive: the implementation of community notes hasn’t been completely unenjoyable. As emerging stories unfold, it’s beneficial to have a touch of extra context adjacent to the original post.

However, I may have yet another reason to permanently depart, as X is piloting a scheme that allows AI chatbots to create community notes. Starting July 1, users can sign up for access to the AI Note Writer API to propose their AI chatbots that can suggest community notes. Like the notes penned by humans, these AI-generated notes will start surfacing later this month if deemed helpful by users across various perspectives.

This pilot initiative aims to enhance both the speed and scale of Community Notes while incorporating community feedback to refine AI agents for delivering more accurate, less biased, and broadly helpful information.

Related Articles:

• The LA Times gives mealy-mouthed AI the last word on its opinion pieces
• OpenAI boss Sam Altman thinks young people turning to chatbots for life advice is ‘cool’

It’s important to note that this announcement comes after a reported 50% drop in the number of community notes created since January. An X spokesperson suggested that part of this decline is due to the natural ebb and flow of contentious topics, like the U.S. elections. Anyone who has recently engaged on X can attest that nothing sparks dialogue quite like ‘controversial topics’.

In consideration of the AI note writer announcement, my foremost concern is… did X forget the well-documented issue of AI hallucinations? You know, the phenomenon where large language models tend to fabricate information while striving to predict the next most likely word, rather than generating factual content?

To be fair to X, there won’t be an unrestricted use of AI. AI notes will be plainly marked for users, and will also be held to the same standards as human notes, with an open scoring algorithm that identifies notes deemed beneficial by a variety of perspectives. However, similar to human-written community notes, there remains the risk of disinformation gaining heightened exposure through vote manipulation.

This development feels reminiscent of the ‘Move fast and break things’ approach; in fact, it’s noteworthy that Meta recently eliminated all of its third-party fact-checkers in favor of its own variant of a community notes system earlier this year. Managing extensive platforms like Facebook and Instagram is a tremendous responsibility and likely even more expensive to execute adequately—delegating this task to a network of volunteers is a pragmatic financial decision.

Nonetheless, it stands to reason that there are various reasons why major tech companies like Meta are championing ‘free speech’ while evading the responsibility of effectively moderating their platforms. Notably, Epic CEO Tim Sweeney recently criticized tech leaders for ‘pretending to be Republicans’ to curry favor with the new administration, allegedly to sidestep antitrust regulations.