THE ONE IMPORTANT STORY

🤖Elon Musk’s xAI rolls out its own AI model

In 2013, when Google acquired DeepMind—the company responsible for AlphaGo, an AI that defeated the world's top Go players— Elon Musk freaked out. Earlier, during a friendly chat, Larry Page, the co-founder of Google called him a “specist” for worrying about AI taking over, and told him “if machines surpassed humans in intelligence it’s simply the next stage of evolution.”

Elon joined forces with Sam Altman to establish OpenAI, a nonprofit dedicated to researching AI in a way that ensures it does not get out of hand and its benefits are distributed globally.

However Musk left OpenAI in 2018 and Sam created a for-profit arm to fund their research. They went on to create ChatGPT.

Alarmed by the capabilities of the chatbots that were released, Musk decided in July to create xAI, his own AI company.

Late to the party: Despite being an early backer of OpenAI, Musk is considered late to the game with his new venture, xAI. Nevertheless, the company managed to attract some of the top talent in the field, poaching them from Google’s DeepMind team and academic powerhouses like the University of Toronto.

After just two months of training, xAI released its first AI model, Grok to a select few beta-testers.

Grok, inspired by "The Hitchhiker’s Guide to the Galaxy," is designed to answer questions with a bit of wit. It appears to be doing just that: early testers posted screenshots where Grok compares API scaling to a never-ending orgy, trolls the user for asking how to make cocaine, and makes fun of SBF facing 110 years in prison.

The company claims that it will also answer spicy questions that are rejected by most other AI systems and aims to be less biased, seeking the utmost truth.

While it may appear like your average chatbot, there are a couple of fun twists. You can do multitasking, prompting Grok with different questions in multiple windows simultaneously, you can engage in branch conversations going down and exploring multiple paths in a conversation, open code snippets in VS Code editor, and edit Grok’s responses, which it will analyze and adjust its style accordingly. This last one is awesome for role plays and creative tasks. You can also switch Grok to ‘Fun mode’ if you want to get more bizarre answers.

The most unique feature of the model is it’s real-time knowledge of the world via the 𝕏 platform. But whether Grok's take on the latest happenings is going to be accurate and fair is anybody's guess, considering 𝕏's got its own issues with spreading misinformation.

How does it stack up to other models?

Well, currently, it's just another model with ChatGPT 3.5-level strength. It's still impressive given the short training period and that it was trained on only 33 billion parameters; however, it’s far from the capabilities of OpenAI’s flagship GPT-4, which is already over a year old (and training likely finished quite a bit earlier).

Once it's out of early beta, xAI’s Grok system will be available to all X Premium+ subscribers. Its $16 per month price tag feels a bit hefty compared to ChatGPT Plus's $20 per month, especially considering the number of features and quality offered by the latter. Twitter powerusers might find the deal appealing though.

The main question for xAI isn't about profitability, however. It's whether the company can catch up with those at the bleeding edge, like Anthropic and OpenAI.

ONE MORE THING

Ilya Sutskever: The scientist who is shaping the world through AI

This short documentary was shot with the Chief Scientist of OpenAI between 2016 and 2019, when his team was developing the language model that is now ChatGPT. 

Although the style of the documentary is a tad bit too dramatic. It’s a great window into the mind of the genius at the forefront of developing AI.

⌚ If you have one more minute

• Everything you can do with Copilot, the AI Assistant in Windows 11

• British PM chats with Elon Musk about the future of AI - video

• AI can now turn a single image to 360º scene

• Midjourney gets a new style tuner feature and it puts back the fun into creating images. Here’s how to use it.

• The (unconfirmed) announcement leaks from today’s OpenAI Dev Day - Twitter thread

• Teen boys use AI to make fake nudes of classmates, sparking police probe

AI Art of the Day 🎨

Check out this tweet on how to use Midjourney's new Style Tuner feature to create images with a consistent style.

🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄

That's all for today, folks!

If you enjoyed this, please share this hand-crafted newsletter with a friend. It would make this writer's day!