Elon Musk’s Grok, an AI assistant developed by his company xAI, has been making headlines with its strikingly candid responses, sparking intense discussions about its unusually human-like political opinions. While some users praise its unfiltered approach, others have raised concerns about abusive responses.
The generative artificial intelligence platform’s ability to mimic human-like language patterns, including an explicit or aggressive tone, has left many pondering the profound implications for the future of AI as it evolves.
Isn’t Grok just another AI chatbot?
In the most technical terms, Grok is xAI’s large language model (LLM) that is trained on synthetic datasets using machine learning (ML) techniques. Its website describes it as a “cosmic guide”, but put simply, Grok is a conversational AI chatbot designed to facilitate both serious and lighthearted discussions on Musk’s social media platform X.
Powered by xAI’s Colossus supercomputer, Grok focuses on analyzing real-time data and providing users with relevant insights tailored to their inquiries and interactions.
Unlike other AI-based chatbots such as OpenAI’s ChatGPT, Google’s Gemini and DeepSeek, Grok has access to a vast amount of data, including real-time information from X and a broad range of internet text data up to September 2023, with additional datasets curated by human reviewers.
Also read | Are we on the cusp of AGI, as Musk suggests?
How else is Grok different from the other AI chatbots?
After the February launch of xAI’s Grok 3 series of models and claims that it could be a replacement for Google’s search engine, users have been engaging with it a lot more.
Users are experimenting with Grok’s new features such as Think Mode, which provides step-by-step reasoning for users, Big Brain Mode, which uses more computational resources to handle complex tasks, and Personality Mode, which lets users try various preset personalities such as ‘romantic’ and ‘conspiracy’.
Is Grok really “scary smart”?
On subjects such as politics, cinema, and celebrities, Grok has left some users bedazzled by its opinionated responses. In other instances, Grok has used regional slang and abuses. These instances only reinforce the label that Musk applied to the AI model while launching it: “scary smart”.
Also read | The rise of ‘agentic AI’: Why data scientists and software developers should be worried
It’s also a matter of concern how AI has advanced to the level of mimicking humans. This can lead to various problems, including job losses and cyber security issues. Data from connectivity cloud company Cloudflare Inc.’s study reveals that 93% of Indian businesses are concerned about AI amplifying the severity and sophistication of data breaches.