X is set to release the next generation of its artificial intelligence tool, Grok, this summer. Elon Musk announced in a late-night tweet that Grok 2 will arrive in August. He acknowledged the challenges of purging large language models (LLMs) from Internet training data but claimed that Grok 2 would make significant improvements in this area.
Musk didn’t provide extensive details, though he had mentioned in March that Grok 2 “should exceed current AI on all metrics.” Despite this optimistic projection, it appears Grok 2’s reign will be brief. Later the same evening, Musk tweeted that Grok 3 would be released by the end of the year. He noted that Grok 3 would undergo training on 100,000 Nvidia H100 GPUs and suggested it “should be really something special.”
Grok, introduced in November as X’s answer to ChatGPT, had its 1.5 version debut in April. The chatbot is currently available exclusively to Premium subscribers on X. Additionally, Musk’s xAI startup is gearing up to build what could become the world’s largest supercomputer in Memphis, Tennessee, to handle Grok’s training.
Like other chatbots, Grok has faced reliability issues. Despite its aim to counteract what Musk describes as the ChatGPT “propaganda machine,” Grok has spread false news stories, including an erroneous claim about Indian Prime Minister Narendra Modi’s election status before voting had even started.
In an attempt to differentiate, X added a “fun mode” to Grok, producing more satirical and juvenile takes on news, which might not be ideal during an election year.
Musk’s ambitious plans for Grok, including rapid updates and extensive computational power, suggest a bold future for the AI tool, though its impact remains to be seen.
