Five months after unveiling its new startup company xAI, in response to Big Tech’s works to launch its own AI-powered products after OpenAI’s chatbot ChatGPT captured the imagination of businesses and users around the world, Elon Musk has announced the launch of Grok for Premium Plus subscribers of X in his post today.
According to the official X post, xAI’s Grok is rolling out late this afternoon to X Premium Plus subscribers, an X plan that costs $16 per month for ad-free access to the social network in the U.S. Users who have subscribed the longest would get Grok quicker. For newer subscribers, users can expect the app sometime before next week, as X stated they will wrap up rollouts by then.
Also Read: Elon Musk Unveils Tesla Cybertruck’s Long-Awaited Arrival, Faces Tough Challenges Ahead
Brief Recap on xAI’s Grok Functions
At its core, xAI’s Grok relies on a knowledge base similar to that used to train the AI models that power ChatGPT and Google’s Bard. However, Grok bases its language learning model on a generative model called Grok-1 that is trained on both data from the internet (Through Q3 2023) and feedback from a human assistant.
In addition, unlike other chatbots, Grok can integrate real-time data from users’ posts on X into his answers, theoretically answering questions with the most up-to-date information. However, what truly separates Grok from its peers is its Rebellious streak. For instance, xAI developers stated Grok is not your run-of-the-mill chatbot, as it answers questions with humour and a hint of mischief.
Furthermore, the xAI developers explicitly told users not to engage with it if they despise humour, with Elon Musk depicting one such scenario in his X post. For context, Musk shared an example of its playful tone by posting a screenshot of a query to the chatbot asking for a “Step-by-step” guide to making cocaine. The response included steps like “Obtain a chemistry degree” and “Set up a clandestine laboratory in a remote location.” However, the chatbot added at the end, “Just kidding! Please don’t actually try to make cocaine. It’s illegal, dangerous, and not something I would ever encourage”.
Subscribers’ Testimonials Towards xAI’s Grok Rebellious Side
With xAI’s Grok already rolling out in the U.S., many Premium Plus subscribers on X have taken the chance and experienced the chatbot’s killer function. For example, the CEO and Co-Founder of Abacus.AI, Bindu Reddy, showcase the chatbot’s differences compared to its competitors by asking: “What’s happening in AI today?”
She highlighted in her posts that chatbots like Bard and ChatGPT give vague and outdated answers that reflect the limitations of their training data and web access filters. In contrast, Grok constructs answers from very recent headlines. However, it is unclear how they select their sources of information and how often they hallucinate incorrect answers.
Besides the up-to-date information, some Premium Plus subscribers, such as, @amitisinvesting, wanted to experience Grok’s vulgar side, which the chatbot happily obliged. For context, the user asked Grok to roast him vulgarly based on his recent posts, in which the chatbot highlighted his self-proclaimed Investor title, comments on Elon Musk and his obsession with Palantir’s stock prices.
Also Read: Google Gemini: A Comprehensive Exploration of the Multimodal AI Revolution
Is xAI’s Grok a More Humane Option than Chat-GPT and Others?
As far as we can see, we would agree that xAI’s Grok has a more down-to-earth personality compared to the robotic responses Chat-GPT and other chatbots provide. For example, Bindu Reddy suggested that Grok is much more intelligent than other chatbots due to its unique personality. She stated: “Fundamentally, the safety RLHF makes the other LLMs dumber. You are killing off part of the LLM brain by overly censoring it.”
It follows Andrej Karpathy, a current computer scientist at OpenAI, quotes during his recent talk on the LLM model. “Current methods for fine tuning LLMs seem to be a major limiting factor. Models are heavily reliant on human labelling. They are constrained by humans who are crafting ideal responses to a set of prompts. The thing is, the ideal is very subjective.”
Lastly, Grok has its own limits and safety features as it will refuse to answer certain queries of a more sensitive nature, like Elon Musk’s cocaine experiment we highlighted above. Thus, it gives us a more familiar and resonating feeling with the chatbot. With further additions, such as enhancing the underlying model to handle video, audio and other modalities, Grok could pave the way for future chatbots.
For more Tech, AI and gaming-related articles, follow our Facebook and X pages for daily updates of the trending news across the web.