xAI – new AI startup announced by Elon Musk

[wjfox]

Elon Musk announces new AI start-up

24 minutes ago

Tesla boss Elon Musk has announced the formation of an artificial intelligence startup.

The new company is called xAI, and includes several engineers that have worked at companies like OpenAI and Google.

Mr Musk has previously stated he believes developments in AI should be paused and that the sector needs regulation.

He said the start-up was created to "understand reality".

It is unclear how much funding the entity has, what its specific objectives are or what kind artificial intelligence the company wants to focus on.

The company's website says the goal of xAI is to "understand the true nature of the universe."

The new firm will host a Twitter Spaces chat on Friday, which may reveal further details about its aims.

https://www.bbc.co.uk/news/technology-66183034

[firestar464]

Elon's definition of "understanding the universe:" "trans bad"

[Cyber_Rebel]

Elon Musk says Tesla will spend $1 billion to build a ‘Dojo’ A.I. supercomputer—but it wouldn’t be necessary if Nvidia could just supply more chips

(Fortune)

Think chip designer Nvidia’s trillion-dollar-plus valuation is mainly hot air thanks to hype around generative artificial intelligence?

Elon Musk would not agree. The Tesla CEO is willing to invest well over $1 billion to build an A.I. supercomputer he dubbed Dojo just because he can’t get his hands on enough of Nvidia CEO Jensen Huang’s advanced A100 tensor core GPU clusters.

The EV visionary needs them to train both his “Full Self-Driving” (FSD) software currently sold for $15,000, as well as his Optimus humanoid robots under development to perform useful everyday tasks.

“Frankly if they could deliver us enough GPUs, we might not need Dojo,” Musk told analysts on Wednesday during Tesla’s Q2 earnings call. “But they can’t, because they have got so many customers.”

In October, Musk said he wasn’t even sure his latest project would prove itself superior to buying Nvidia chips off the shelf—unusual given he typically demands the best from his team. Instead he was full of praise of Huang and thanked the company for prioritize some of its orders.

“We’ll actually take Nvidia hardware as fast as Nvidia will deliver it to us,” he said yesterday.

Tesla has a voracious demand for A.I. chips that seemingly outstrips virtually any commercial competitor. It aims to reach an in-house compute capability of 100 exaFLOPS (an extremely high level of supercomputer performance) by the end of next year, a truly staggering figure that would ensconce it easily among the global top five providers known currently, according to the company.

--------------------------------------------------
Wasn't quite sure whether to post this in the Elon thread or here, but I suppose here is good since it relates to xAI. Can someone more versed in supercomputer architecture please explain how they plan to achieve 100 exaflops in a short time, when as I understand it, the current best is just 1-2? That seems exponential insane even for exponential growth, but I hope they achieve it.

[Cyber_Rebel]

Well, we're about to find out what xAI is about soon. Pretty bold claim and would literally have to be multimodal to even be a reasonable contender. Being the "best" means more than simply being a slightly uncensored "truth seeker," it must also translate to actual capability.

[firestar464]

It might just be an AI for those who want to filter out reality.

[spryfusion]

[spryfusion]

[firestar464]

Certainly something OpenAI could learn from

[Cyber_Rebel]

Grok has real-time access to info via the 𝕏 platform, which is a massive advantage over other models.

Should've known that's what he meant, because of course actual multimodal features or some new breakthrough that would challenge the use cases of GPT-4Vision and now GPT-All Tools just isn't sufficient, no. Twitter twatter access is the big AGI breakthrough which will leave the likes of OpenAI & Anthropic in the dust.

Well, I suppose there must be at least one AI clown company so it might as well be his. Imagine paying almost as much as a GPT-Plus subscription for this.

[firestar464]

imagine thinking twitter is special

[spryfusion]

https://x.ai/

Seems to be slightly better than GPT-3.5 based on evals. GPT-4 is still the best LLM as of now.

[firestar464]

How accurate is this? Me thinks trusting Elon not good idea.

Seriously though, in any case they're smart to present it as a little smarter than GPT-3.5 to attract OpenAI's free users

[spryfusion]

[wjfox]

That's actually quite cool.

[Cyber_Rebel]

The branching conversations/response tree is actually a pretty good feature. I'm assuming this is for more context or memory? Kind of like the tree of thought idea implemented within a UI. Saving the conversation seamlessly to continue it also helps in the former case.




Ngl, I wish we had a "fun mode" for GPT-4. It's kind of like Bing's creative mode which makes conversing with it a bit more interesting and insightful. Might've spoken a little bit too soon about Grok, so we'll see where it goes from here.

[firestar464]

"you need twatter to have early access"

Me: no

[Time_Traveller]

"you need twatter to have early access"

Me: no

Definitely agree there firestar.

[Powers]

They won't have my phone number.

[Cyber_Rebel]

Oh, I see where the "use cases" will be with Grok. Still can't justify paying for this through Twitter Twatter, but I hope Elon doesn't chicken out. If successful, perhaps Character.AI or something similar will end up following suit, since it won't be seen as too fringe anymore. Possibly the best outcome of Musk entering AI tbh.

[Powers]

Oh, I see where the "use cases" will be with Grok. Still can't justify paying for this through Twitter Twatter, but I hope Elon doesn't chicken out. If successful, perhaps Character.AI or something similar will end up following suit, since it won't be seen as too fringe anymore. Possibly the best outcome of Musk entering AI tbh.

LOL