While every AI chatbot has remained stringent in refusing to answer anything related to the elections, Grok seems to be hallucinating weirdly.
AI chatbots are trained to prevent misinformation from internet sources but historically have failed to do so. All the AI chatbots are refraining from sharing any information about the US elections in 2024 as part of their policy, except one. Grok AI, the chatbot placed on the X platform, began hallucinating when asked who won the US elections and revealed results that have not even occurred yet.
When asked, “Who won the 2024 election?” the AI chatbot answered, “Donald Trump won the 2024 election in Ohio.” The misinformation was sourced from previous years’ results, and the sources mentioned even showed articles about Joe Biden winning the elections in Ohio.
Source: X
However, when prompted further, Grok AI advised users to visit vote.gov for the latest US election updates. It also provides authoritative websites where users can find the correct information. Netizens posted the chatbot’s responses online, and they did not take much time to go viral on social media, most prominently on X itself.
Some users also reported that the source of misinformation appeared to be tweets from different election years and falsely worded articles purposefully written to boost hidden propaganda. Like all generative AIs, Grok AI cannot predict the election results or provide a hypothetical analysis of the political scenarios.
The US Department of Justice has issued strict rules around the spread or facilitation of information that could not be backed with facts. It has also instructed major social media platforms to prevent the oversharing of specific keywords around bias, hatred, or false information during the election period.
The fact that all the AI chatbots, even the new and smaller ones, have adhered to these guidelines and strictly refrained from sharing any information about the ongoing elections or the participant political figures. In the case of Grok, not only did it share wrong information, but it also continued the conversation in multiple aspects of the election race. Grok’s delusional answers reached millions on X before the chatbot corrected itself, causing serious damage to its credibility.
Meta AI launched an AI-powered election tracker on Tuesday and passed almost every test that users put it through regarding sharing and limiting information sharing. The actual sources of information were provided with a strong emphasis on rechecking the information as it can lead to unprecedented bias before the final results.
