Recently, many companies, including social media platforms, have been working on artificial intelligence. X, formerly Twitter, launched its chat bot Grok in recent months. Elon Musk, the CEO of the social media platform, said that this bot was fun and even said that those who hate humor should not use it. But according to recent developments, Grok, while fun, doesn’t seem to be very good at understanding humor-related content. Here’s why…
Grok mistook a humorous content for real news
Accessible to users with a Premium subscription to X (Twitter), Grok is a tool that can answer almost any question you want to ask it. It meets its information needs from the social media platform. So you can ask it a question about a popular topic.
Grok needs to learn first to provide information on any topic. In order to learn, he turns to the most popular, that is, the most liked posts. But what would happen if these posts contained humorous content? This happened.
In the last hours, a 4.7 magnitude earthquake occurred in New York, one of the largest cities in the United States. Reportedly 10 kilometers deep, the tremor was felt for several seconds.
After the earthquake, an X (Twitter) user posted that the mayor of the city, Eric Adams, had deployed the New York Police Department (NYPD) to prevent earthquakes, was considering using “robo-cops” and ordered “every cop to shoot and destroy earthquakes”.
As you can imagine, this post, shared for humor, received a lot of likes. It was then revealed that Grok had mistaken the content for real news and summarized the information and presented it to users. Thus, it was revealed that the chat bot was not very good at distinguishing between real and joking or humorous content.
{{user}} {{datetime}}
{{text}}