Wow. The currently best ranked consumer AI model has just acquired the most complete corpus of scaled, real time information on the Internet. This data will be a part of the pretraining to make the models xAI makes even more differentiated. This is a smart move in a moment when other model makers are caught up and slowed down in copyright lawsuits (OpenAI) for training data or pre-training quality (Meta) itself.

Elon Musk @elonmusk

6 months ago

Wow. The currently best ranked consumer AI model has just acquired the most complete corpus of scaled, real time information on the Internet. This data will be a part of the pretraining to make the models xAI makes even more differentiated. This is a smart move in a moment when other model makers are caught up and slowed down in copyright lawsuits (OpenAI) for training data or pre-training quality (Meta) itself.

16K 31K 235K 59.9M 15K

@chamath So @grok will now be as smart as the people posting on @X.

@chamath This was clear from the beginning. The $44b valuation made sense with something like this in mind. He pretty much telegraphed it.

FischerKing @FischerKing64

3 years ago

@chamath This was clear from the beginning. The $44b valuation made sense with something like this in mind. He pretty much telegraphed it.

15 24 289 0 20

@chamath You wrote this with AI didn’t you

Training on real-world data without transparency or diversity of sources risks turning these systems into echo amplifiers, not truth seekers. Add in the fact that it’s one entity controlling both the data and the model—and you’re not building intelligence, you’re bottling perspective. Performance is impressive. Power concentration is the real problem.

@chamath @elonmusk just out-accelerated sly @sama big time

@chamath Synergy value is through the roof. How will others even compete? Just like Tesla’s video library advantage with FSD. Pretty genius first mover advantages by Elon.

@chamath Now Tesla should acquire xAI and bring this full circle.. Do it @elonmusk

@chamath The "acquisition" may be a smart move, but using X posts to train xAI would be a very bad idea. I didn't see @elonmusk mention that X data would be used in training xAI.

@chamath They were already using X data obviously

@chamath I think he is good on anti trust. 😉 They fit very tightly.

@chamath I wonder if any other LLMs are going to try a similar path with social media. And agree, this is HUGE!

@chamath Hasn’t he stated xAI has already been using X realtime data? What will change?

@chamath Do you think this has a material impact on either business?

@chamath Mutations are happening faster than we can think “Acceleration is accelerating” e/acc is on the move @GillVerd

@chamath There’s tremendous potential for growth here. People wrote off “data” as a commodity, but in the coming 1-2 years I think we will realize that not all data is created equal.

@chamath "Information"

@chamath Twitter was the training dataset.

@chamath How realistic is copyright enforcement in the age AI?

@chamath

@chamath I find it hard to believe that wasn't already the case, wasn't that already its edge. been using Grok for recent news research since day 1

@chamath In the future, no AI company can survive without a strong data platform. This is more evident now than ever before @X and @Xai marriage is an example @chamath brought this topic before on @theallinpod

@chamath This was a master move ♟️

@chamath @grok explain this

@chamath i have to imagine it was already apart of the data

@chamath Yes, a very big deal

@chamath Yeah this is pretty dope, smart move.

@chamath Or folding a bankrupt company into a better capitalized one. Just like bankrupt solar city.

@chamath Chamath, by what reasonable metric is Grok the best ranked consumer AI model, exactly?