Elon Musk has unveiled a new AI chatbot called Grok for his social media platform X (formerly Twitter), claiming even the beta version has more current information than rival chatbot OpenAI’s ChatGPT.
Musk’s xAI company developed the chatbot in two months and warned it was still a ‘very early beta product.’
Grok is currently available for testing with a select group of US based users. Grok AI assistant will be provided as part of X’s $16 per month Premium+ service once it leaves beta.
Grok was developed using the information posted to X and therefore claims to have faster access to real-time developments, xAI’s website said.
Musk noted Grok would also answer questions other models are designed to reject. During a demonstration of Grok, a user requested a step-by-step guide for making cocaine for which the chatbot provided generalised instructions before warning: “this guide is just for educational purposes.”
Jake Moore, Global Cybersecurity Advisor at Slovakian cybersecurity firm, ESET, said that while blocking criminal requests can help prevent low-level criminals from accessing resources, “the more advanced threat actor has been known to circumnavigate these resistances but enabling them by default may just direct scammers and cybercriminals to this latest AI chat model.
”However, this is arguably similar to inputting rogue search requests into popular search engines which already offer links to blogs with criminal content even though giants like Google attempt to keep malicious content from being linked to their site.”
The term ‘grokking,’ meaning to understand (something) intuitively or by empathy, originated in Robert A. Heinlein’s 1961 novel Stranger in a Strange Land. xAI said Grok was modelled on Hitchhiker’s Guide to the Galaxy, a comedy sci-fi series by Douglas Adams.
At the UK’s AI summit last week (1 Nov), Musk acknowledged the disruptive, and potentially dangerous, aspects of AI, warning “there will come a point where no job is needed”.
Musk has been a long-standing advocate of AI technology and played a co-founding role in OpenAI, which created ChatGPT last November.
Since then, Meta launched Llama in February, followed by Google’s AI model, Bard, in March.
The narrative around the existential threat of AI dominated last week’s AI safety summit, with Elon Musk among those voicing concerns, according to GlobalData principal analyst Laura Petrone. “However, this narrative fails to adequately acknowledge the short-term risks likely to dog AI chatbots, including Grok. Grok has already proved to have the same shortcomings as other generative AI systems like ChatGPT, in particular when it comes to hallucinations and inaccuracies,” she said.
Petrone believes that X’s trouble with content moderation could also impact Grok, as the AI chatbot has access to X’s data. “The risk is that Grok would regurgitate the biases and harmful content that have dogged the platform since X’s owner, Elon Musk, decided to relax its moderation policy,” she said.
The fact that Musk believes there is “significant danger in training AI to be politically correct” suggests that an unfiltered free speech approach will also inform Grok, adds Petrone.
GlobalData’s Thematic Intelligence: Artificial Intelligence report estimates that the AI market will grow from $81bn in 2022 to $909bn by 2030, at a compound annual growth rate of 35% between 2022-30. In the coming decade, the country that emerges on top in AI will lead the Fourth Industrial Revolution.
