Elon Musk has done it again, unbalancing an industry, this time with the introduction of the Grok AI. The ChatGPT rival is not the first AI platform to make its debut since the more open acceptance of AI—big names like Claude, Watson, and Bard are a few of the other names we’ve grown familiar with over these last two years. However, Grok by Elon Musk brings an assortment of unique features that might endear the chatbot to its users and also boost the Twitter/X platform that has been gaining notoriety but failing in popularity and use.
Grok AI: Release and Access
Grok by Elon Musk was formally announced through a blog post on 4 November, on xAI’s blog, the company designing the AI. While little is known about the full extent of the Grok AI’s capabilities, the blog post and follow-up tweets by Elon Musk have hinted at the ways this AI chatbot will set itself apart from its competitors.
According to the article, the AI takes inspiration from Douglas Adams’ Hitchhiker’s Guide to the Galaxy, a pioneering work of science fiction literature. The Grok AI name likely also takes inspiration from Robert A. Heinlein’s sci-fi work Strangers in a Strange Land, with “Grok” derived from the language of Martians to mean drink. His recent tweet with just the name of the book alludes to the same. According to Grokcode, this also means “to absorb thoroughly” or understand so well as to become a part of the experience.
A noteworthy detail about the ChatGPT rival is that its development of the AI was quite rapid, with just about 4 months of development and 2 months of training. The large language model (LLM) will continue to scale up significantly with use. The Grok AI is only available through early access to a limited number of verified X users in the U.S. Users who sign up on the Grok waitlist will be able to test the model and provide feedback as necessary. Unfortunately, this will likely never be available for the public to test out freely, and will only be available for those with a Premium+ X account, currently at $16 per month.
Grok by Elon Musk: “Worry about a Terminator Future in Order to Avoid a Terminator Future”
Musk has been surprisingly hesitant about AI and its rapidly advancing capabilities so far, warning against its use after he walked away from OpenAI, a company he co-founded but did not continue to support. Despite his stance that AI could become “one of the biggest threats to humanity”, he set up X.AI Corp to develop artificial intelligence that would be “pro-humanity” due to its curiosity about mankind rather than being based on ethical or moral grounds. If you can’t beat them, join them and do better—this appears to be the aim of Grok by Elon Musk since there is no way to now pause and restrict work on AI and LLM models.
According to The Guardian, the xAI team is led by multiple, highly experienced individuals from DeepMind, Google, and Microsoft. With experience backing the technological capabilities of the Grok AI, there is hope for the AI to catch up to its competitors in a while. Dan Hendrycks, director of the Center for AI Safety (CAIS), is also reported to be an advisor to the xAI team. His is a prominent voice among the naysayers of AI and the concerns around the risk of extinction with continued and unchecked AI growth.
Grok AI: What Sets This ChatGPT Rival Apart From the Rest?
Right off the bat, Grok by Elon Musk inevitably has signs of his involvement in the AI’s brand identity. “Grok is designed to answer questions with a bit of wit and has a rebellious streak, so please don’t use it if you hate humor,” the xAI post announces and from Musk’s follow-up tweets, it is clear that the chatbot is one with a sense of humor, quite different from the more formal approach of most other chatbots. The Grok AI responses appear to be conversational, humorous, vulgar, and sarcastic, traits that the CEO appears thrilled about.
Grok has real-time access to info via the 𝕏 platform, which is a massive advantage over other models.
It’s also based & loves sarcasm. I have no idea who could have guided it this way 🤷♂️ 🤣 pic.twitter.com/e5OwuGvZ3Z
— Elon Musk (@elonmusk) November 4, 2023
While this might make the chatbot more engaging, what stands as the ChatGPT rival’s best feature is its ability to gather data in real-time. Again, while some search engine AI models like Bing already have real-time data reproduction, the Grok AI information bank should do well paired with the Twitter/X platform. The Grok-1 model card page does mention its limitations and inability to search the web independently with additional search tools deployed for the capabilities of the Grok AI. It highlights the need for human review of the data presented by the Grok-1 model. “The model can still hallucinate, despite the access to external information sources,” the website states.
xAI’s elaboration on future research directions for the AI model is also noteworthy. It mentions a focus on adversarial robustness to ensure the vulnerabilities of AI are addressed effectively. The future plans for the platform to engage multimodal audio and visual capabilities are also interesting but with its competitors already making great advancements, there might need to be more built into it for it to stand out.
Benchmarks and Initial Performance Reviews
For now, the xAI team appears satisfied with the Grok-1 performance and they highlight the tests that have been conducted to compare its performance on standard machine learning benchmarks with competitors like GPT 3.5 and LLaMa 2 70B, and its own predecessor model Grok-0. Competitors like Claude 2 and GPT 4 far exceed the performance of the Grok AI, but with only 4 months of development, the AI seems to be holding its own ground.
— Viv (@battleangelviv) November 5, 2023
Grok by Elon Musk is clearly designed for the kind of users the X Premium subscription currently attracts, those with a more edgy approach to all things. Time will tell how the early access program plays out and if the Grok AI is satisfactorily able to draw more users back to the platform on a more permanent basis but for now, news of the AI is limited to the tweets we have so far.