What’s The Story Behind Musk’s Latest AI Called “GROK”

Elon Musk has announced his own large language model to take on the likes of ChatGPT, Claude and Pi.

Developed by X.AI, Musk’s AI company, the large language model is called GROK and is presented as an AI agent designed to answer any question conversationally, with a hint of wit and rebelliousness.

This is how X.AI introduced Grok on the 4th November 2023:

“Grok is an AI modeled after the Hitchhiker’s Guide to the Galaxy, so intended to answer almost anything and, far harder, even suggest what questions to ask!

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!

A unique and fundamental advantage of Grok is that it has real-time knowledge of the world via the 𝕏 platform. It will also answer spicy questions that are rejected by most other AI systems.

Grok is still a very early beta product — the best we could do with 2 months of training — so expect it to improve rapidly with each passing week with your help.

Thank you, the xAI Team*

This sales pitch is enough to get your attention, but be careful, the model is still in its early beta phase after only having two months of training on data.

In truth, it’s far too early to “launch” GROK and says more about Musk’s beef with Sam Altman and OpenAI than it does about his desire to enhance the world of AI. This announcement came at the same time as Altman was leading OpenAI’s first ever developer conference!

The name GROK is a term popular amongst science fiction fans and was coined in Robert Heinlein’s classic 1961 sci-fi novel, ‘A Stranger in a Strange Land’. Although panned by some critics, the novel was an immediate success and the first work of science fiction to make The New York Times Book Review’s best-seller list.

Heinlein’s coining of the fictitious Martian word signifies a profound understanding of something at an intuitive level. Musk’s choice of GROK for his AI model tells us how he plans to position it. The cynic in me thinks this is a case of Twitter deja vue all over again! Over promising on expectation and under delivering on execution!

And given the GROK has been trained using real-time access to Twitter data, Musk has opened GROK up to the criticism that it might not be that intuitive after all!

One of GROK’s key selling points is its access to real-time Twitter data, a feature that no other AI has at the moment, thanks to Musk’s decision to shut off Twitter’s API to other AI companies. While the value of Twitter data can be debated, its potential to provide a real-time representation of society could be a game-changer for GROK, but only as long as Musk can keep Twitter going as a platform of choice for the world’s news.

According to X AI’s internal evaluations, GROK is performing at a level that’s essentially equivalent to GPT-3.5 (the free-to-use version of ChatGPT). However, these assessments have yet to be validated by external parties.

Supercharge Your Career With AI

The Wiser! newsletter is for professionals, creators and solopreneurs who want to master AI to advance their careers.

Join the mailing list, get Wiser! every week, and claim your free download of The Beginner’s Guide To ChatGPT.

➜ Here’s The Thing: Musk’s decision to develop GROK isn’t a sudden move. In fact, it’s deeply rooted in his history with OpenAI, desire to stay at the forefront of the AI agenda, and address his problems with Twitter.

Musk was one of the co-founders and original investors in OpenAI, which was created in 2015 to be an open alternative to what Google was building at the time. Musk, Sam Altman and the OpenAI founders worried that Google was going to corner the market on an AI and they didn’t care about doing it in a responsible and safe way. OpenAI was founded to as a research AI company for the sake of humanity.

However, in 2018, Musk had a disagreement about the future of OpenAI. He failed in his attempt to become CEO of OpenAI and unceremoniously left the AI research company. Shortly afterwards, OpenAI launched the for-profit part of their business.

Since then, Musk has expressed his disapproval with OpenAI’s approach to building AI systems. He’s also been vocal on the risks to humanity posed by the likes of OpenAI. Ironically however, this hasn’t stopped Musk from building his own competing AI company, called X AI. And built his own AI large language model, which is GROK.

While it’s too early to tell if GROK will be a credible and viable competitor to the likes of ChatGPT, Claude, LLaMA and PaLM, it’s clear that Musk’s motivations extend beyond just creating another foundational AI model. He has a personal stake in developing his own AI that aligns with his vision and the needs of his various business initiatives.

He obviously has Tesla, which is a major AI company trying to create autonomous driving. He has Optimus robots that he’s trying to build intelligence into. And he has Starlink which no doubt sits on a ton of data.

He also has Twitter, his $44 billion hobby that he’s trying to figure out what to do with. Musk turned off the API to access Twitter months ago so that other AI companies could stop scraping data off the platform. It’s clear now why he did that because Twitter data is the bedrock for GROK.

But the issue for Musk is also financial. Two weeks ago, an internal valuation put Twitter at less than half what he bought it for a year ago, at around $19 billion. Which, frankly, still seems over valued to me.

Based on current estimates of $2–3 billion annual revenues, the market value of Twitter would be closer to $6–10 billion if you used Facebook’s 3x multiples as a crude reference point.

The point is that one way to make Twitter worth more is to change the way the market values it. Which means redefining Twitter as a 10x AI company, not an 3x advertising-based social media business model.

Is that his mission for Twitter? Maybe, time will tell!

For now, it’s still a guessing game. The true potential in GROK is all speculation even though it has two unique components — Twitter data and Musk’s genes.

All the more reason to approach GROK with caution. Given its short development time, it’s unlikely that GROK has undergone extensive safety testing, called “red teaming.” This raises concerns about how it could be used and what it might say.

Regardless of all that, it seems that GROK’s release is inevitable, although it’s success isn’t!

Get Wiser! every Sunday!

About The Author

Rick Huckstep has worked in technology his entire career, as a corporate sales leader, investor in tech startups and keynote speaker. From his home in Spain, Rick is thought leader in artificial intelligence, emerging technologies and the future of work.

🕸️ Check out Rick’s website

🤔 Join The Mailing List and Get Wiser! every week (and your free eBooks and resources)

📽️ Follow Rick on YouTube

🎙️ Listen to the Big Tech Little Tech podcast

--

--

Rick Huckstep - Making Sense Of Tech

Supercharge your career with AI - 10x your productivity, prospects and wisdom with tips, tricks, tools and insights about AI and emerging technologies