Mags has spoken. Then deleted. Whoopsies.

They’ve gone full retard.

RT @elonmusk:

Grok is still new. Although LLM technology has been around for a while, the xAI team still had to build Grok from scratch very recently. They did so much later than other companies, other AIs had years of head start. LLM technology was not new, so to a certain degree they were re-creating a wheel that had already been invented. What is impressive though is 1) the xAI team is very lean compared to other organizations, 2) they have moved extremely quickly thus far, 3) Grok punches well above its weight when looking at raw # of parameters vs equivalent LLMs, and 4) the xAI team is now breaking into new territory with scaling hardware. This hardware scaling is innovation we have not yet seen from other companies, and I’m sure comes with its own challenges, and they are learning. It does obviously bring new promises and potential new capability to the world of AI. Grok 2 is very good considering the speed at which it was developed and the # of parameters it has vs. competitors. Punches well above its weight class. I largely agree that other AI models have the edge by a good margin. They have woke bs, but otherwise are still more capable. Grok 2 is still good and delivers good quality results, making it high value for the vast majority of people. The difference in Grok 2 vs openAI models probably is only needed by half of hardcore AI users (not including coding). Grok is much cheaper and has good X integration and webpage source citing. This will only get better, as well as when Grok does more things behind the scenes with the X platform. The value proposition will continue to rise. Grok 3 I get the feeling that xAI had not yet implemented COT (Chain of Thought) / “Reasoning” as of a month ago, and that Grok 3 was going to be released as a standard LLM. Deepseek R1 and o3 implement reasoning, and the xAI team learned about this new method shortly before their intended release of Grok 3. Obviously DeepSeek R1 released its white papers, and o3 released immediately after and already had reasoning capability. I get the feeling the xAI team was blindsided by this to some degree, and is rushing to implement some form of Reasoning capability on top of Grok 3 right now. I expect Grok 3 to be very good. I expect it will close the gap even further. I’m still impressed how far they have come with such a lean team. That said, I do think Grok 3 will be a competitor to o3/etc rather than blowing those AI models out of the water. It may even still be a little behind. I’m not terribly concerned with whatever gap there is, which I expect to be smaller than it is now. As the xAI team continues to scale their hardware and grow their expertise into more and more Grok models, I very much expect Grok to pull away from the competition. I don’t believe we have hit escape velocity yet. Once the team perfects any challenges with hardware scaling, further improves their reasoning algos (in addition to any modifications they need to make at the beginning of training to more deeply implement reasoning capability before training starts), Grok will begin to outstrip the competition by orders of magnitude. Colossus is supposed to have 1 million GPUs by end of 2026. It looks like Grok 7 or 8 will be trained on this, and have hardware scaling perfected by then, in addition to most other AI capabilities. That is the Grok model I am most excited about. I think 2027 is going to be an incredible year for AI across the board. When the Grok team is on par with capabilities with other companies, their hardware scaling will absolutely blow the competition out of the water. Until then, I think we will be seeing large jumps in AI capability, but 2027 will bring a new level of AI intelligence we have not yet seen. Grok escape velocity, 2027. All of my opinions are based on anecdotal rumblings, not hard data. I have not used most of these AI models in a hardcore way, I am just summarizing what I have heard and my gut feeling about the current state of things.

Is there a place to give feedback on the Grok UI / formatting? I have feedback about the desktop UI.

RT @alpha5tate: Phase I, Phase II, Phase III: A Traitor’s Justice

RT @DOGE: DOGE website is live! Initial site: 1. X feed posts 2. Consolidated government org chart - enormous ma…

@alpha5tate I can’t wait until these guys go to jail 😂

Similar to stonk shills, all they have left is character assassination. No substance to go on.

Moon. Maybe tinfoil maybe not, but it’s fun, just a fyi on his post

Tell me you want to steer the narrative away from Q without telling me. RC points to Trump. Trump and Elon are rooting out corruption. Merger needs Kash Patel. Rooting out naked shorting crime ties in with all of this. It’s … interesting at this point not to be curious 🧐 Q

You can check out logins in your settings.

RT @elonmusk: BINGO

@Captaintrips333 @Azkyrie @therealdubbleOO @edwinbarnesc @CocoMoonLe @Life_of_Duyen lol point taken

Trump: No more blood. Ophelia: Blood is on your hands. As an American, I would like to say: Your welcome that Trump is bringing peace, less people are dying, and your/our government’s money laundering is coming to an end. The alternative of course is that the US gov’t stops sending Ukraine weapons, and you negotiate with Putin 1 v 1 without American support. Then you would really get bent over. The Biden regime (CIA) shut down peace agreement offers multiple times.

@RepCasar Hey Greg I’m sure you have salary to spare for the social security benefits. Maybe some stock tips for the elderly too. 100% returns a year would really help the common man.

Overheard today: “Yeah it’s really scary Elon controls all those satellites.” 😑

@alpha5tate My guess as to the order of release, for entertainment ; JFK -> UFO -> Epstein -> 9/11 I hope they release the full details of the tech used on 9/11. And the scud missile

@Rep_Stansbury Which vital programs? Please list them out, with dollar amounts. It’s available online now. Easy

Propping up a fake president for 4 years who is mentally gone is also a crime. We can ask Elon to remove the hat, problem solved. Democrats had 4 years to [legally] swap in Kamala for Joe, but ya didn’t … and lied to the people the whole time.