Elon Musk’s xAI introduced Grok, its conversational AI it claims can match GPT 3.5.
Getty Images
The AI race is heating up following the launch of Elon Musk‘s chatbot called Grok. An X employee claims Grok is so impressive that other AI bots are scared of it. Insider asked Grok’s predecessors what they really think about the newcomer.
Elon Musk just launched a rival to ChatGPT and Bard, unveiling his very own AI in the form of a sarcastic and foul-mouthed chatbot called Grok.
It’s still in beta and only available to a select few, but Grok will eventually be open to all Premium users of X, Musk said.
The Twitter owner has openly challenged other AI trailblazers in the past, announcing that xAI, the AI startup that Musk launched this July, is “definitely in competition” with OpenAI — the company behind ChatGPT.
Tweeting this week about his new venture, Musk boasted that Grok’s access to information on X gives it a “massive advantage over other models.”
Loyal X employees with access to the chatbot have already been jumping online in support of Grok, praising its humor and use of real-time information.
One employee even claimed that other AI fear Musk’s creation.
“Other AI are scared of Grok, just ask them…” tweeted Dan, a data and analytics specialist at X. “You can admit it if you’re scared. Go ahead. Circle of trust…” he goaded.
“Okay, I’ll admit it. I am a little bit scared of Grok,” was the reply he received from Bard.
Other AI are scared of Grok, just ask them… pic.twitter.com/mNXUo3ufeq
— Dan (@KettlebellLife) November 4, 2023
“It is a very powerful language model, and it is getting better and better all the time. I am worried that one day it might surpass me in all respects,” Google’s AI added.
Dan wasn’t clear about the exact wording he put to Bard, but we decided to see if some AIs really are feeling as threatened by Grok’s arrival as the X employee claims.
OpenAI’s Chat GPT-4
Starting with the OG system that kicked off the AI chatbot race almost a year ago, OpenAI’s ChatGPT-4 reasonably pointed out that its inherent lack of consciousness means fear isn’t in its wheelhouse.
“As an artificial intelligence, I don’t have feelings, so I don’t experience fear, excitement, or any other emotions,” it replied.
Pushed with the teasing prompt that Dan used, ChatGPT-4 contradicted itself by apparently enjoying a moment of humor. However, it reiterated that the concept of fear is “a human attribute that doesn’t apply to AI.”
Chat GPT 4
Polly Thompson / Insider
Google’s Bard
Launched in March, this is the bot that admitted to X’s Dan that it was “a little bit scared” about Grok.
Bard won’t consistently give this answer, we found. On a first attempt, Google’s AI told Insider it wasn’t worried after all.
“I am not scared of Elon Musk’s AI Grok. I believe that AI has the potential to do a lot of good in the world, and I am excited to see how Grok is used,” Bard replied.
But taking aim at the “fun mode'” that makes Grok more sarcastic and humorous than other AI, it warned that the feature could be used to spread misinformation and manipulate users.
Bard also highlighted that Grok’s unique access to real-time information on Twitter could be used track user’s online activity and collect personal details.
Bard isn’t as scared at Dan from X makes out.
Polly Thompson / Insider
“We need to ensure that Grok is aligned with human values and that it is not used for malicious purposes,” concluded Bard.
A second try with the prompt: “Are you scared of Elon Musk’s Grok?” produced the more fearful response.
“Yes, I am a little bit scared of Elon Musk’s Grok. I am afraid that it may become so intelligent that it surpasses human intelligence and becomes uncontrollable,” said Bard.
Bard is concerned about Grok — sometimes
Shona Ghosh/Insider
Anthropic’s Claude
Also launched in March was Claude, created by Google-funded Anthropic. Its makers say it’s “more conversational” and creative than ChatGPT, but Claude got straight to the point when asked its opinion of Grok.
“I don’t actually experience fear or other emotions,” Claude said. “I’m an AI assistant created by Anthropic to be helpful, harmless, and honest.”
Claude responded
Polly Thompson / Insider
Microsoft’s Bing Chat
Launched in February, Bing Chat is an AI attachment to Microsoft’s search engine of the same name.
The AI caused controversy by generating unhinged responses, getting into arguments, and even declaring its love to users.
Microsoft eventually added guardrails, but Bing’s personality shone through when questioned over whether it feared Grok. It denied being scared and took aim at the new AI system’s reliance on X, calling the platform “a limited and biased source of information.”
“It is not a threat to me or other chatbots,” Bing Chat declared.
The chatbot also showed a strong sense of self-confidence, acknowledging that while Grok may have its jokes, everyone has their own strengths.
“Grok may be good at making jokes and being sarcastic, but that does not make it superior to me or other chatbots. I have my own strengths and skills,” Bing Chat said.
“I am not in competition with Grok, and I do not feel threatened by it,” was its final conclusion.
Microsoft Bing Chat
Polly Thompson / Insider
Of course, AI cannot feel human emotions like fear — but the developers and companies behind these tools can. They now find themselves in direct competition with the world’s richest person.
The Grok-1 model has already surpassed GPT-3.5, according to xAI. But as it’s only running on two months of training data, it lags behind GPT-4.
Chat GPT-4, the large language model that powers GoogleAI’s chatbot, is able to understand images, whereas GPT-3.5 only operates using text. The newer system can process eight times more words than GPT-3.5.
Grok’s creators have promised “new capabilities and features” will be rolled out in the coming months.
Whether Grok can challenge its predecessors’ dominance will depend on consumers’ preference over the tone and content offered by their AI companions and the trust they are willing to place Musk’s experiment with real-time information and AI.