Grok, the chatbot constructed by Elon Musk’s xAI and popularized on his social media platform X, seems to have repeatedly unfold misinformation about immediately’s mass taking pictures at Bondi Seaside in Australia.
Gizmodo pointed to plenty of posts the place Grok misidentified the bystander — 43-year-old Ahmed al Ahmed — who disarmed one of many gunmen, and the place it questioned the authenticity of movies and images capturing al Ahmed’s actions.
In a single publish, the chatbot misidentified the person in a photograph as an Israeli hostage, and in one other publish introduced up irrelevant details about the Israeli military’s therapy of Palestinians. In one other publish, it claimed a “43-year-old IT skilled and senior options architect” named Edward Crabtree was the one who truly disarmed a gunman.
Grok does seem like fixing a few of its errors. Not less than one publish that reportedly claimed a video of the taking pictures truly confirmed Cyclone Alfred has been corrected “upon reevaluation.”
And the chatbot subsequently acknowledged al Ahmed’s id, writing that the “misunderstanding arises from viral posts that mistakenly recognized him as Edward Crabtree, probably because of a reporting error or a joke referencing a fictional character.” (The article in query appeared on a largely non-functional information web site which may be AI-generated.)
