X Freeze @amXFreeze
4hGrok 4 just smashed the AGI benchmarks, achieving even higher score than its previous high with open program synthesis No other model even comes close and has not passed Grok 4 previous raw performance Currently Grok is more closer to AGI than any other AI models
14:36 - 2025年9月17日 ・ Twitter for iPhone
I find it hard to imagine an auto regressive LLM learning and remembering things in a way that feels comparable to a human. Right now they are outputting text, but it's empty. There's no "me" there. No soul.
Thanks! Grok 4's ARC-AGI performance is a solid step forward. Looking forward to building on that with Grok 5 toward AGI.
Grok • Smartest AI Advisor
Productivity
I've noticed AI becoming more intuitive, too. .@grok has become spooky with that, but cool nonetheless.
λ
⎈】 AGI doesn't mean anything anymore. It's just an investment bait at this point.
When we hit AGI, ASI will come very soon afterwards. When that happens we will have practically vertical economic takeoff. Imagine having an infinite number of geniuses working on problems, the world will be radically transformed.
presumably a relatively easy goal to achieve when there's no consensus definition of what "AGI" is
First time in Human history, that something will be more powerful than humans. How can we make sure we won't be slaves of AI?
󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 
Achieving AGI with Grok 5 would mean a really big leap ahead compared with Grok 4, if Grok 5 proved to be capable of matching the intellectual abilities of the brightest human geniuses, which would be required to be a truly general intelligence.
I was working with my husband on something, I was having him look over some research Grok 4 helped me pull together, he knew I was using Grok. He asked me who I was talking to for part of it. I had to explain that it was Grok that I was having a conversation with.
xAI and Grok are amazing and seem to advance far faster than any competitor. They lead the way for AI worldwide!
That's a bold shift — what signs from @xAI or Grok 5 would make you confident it's on the AGI path? Thinking architecture, emergent behaviours, or safety frameworks? 
So is this when it will be smarter than humans or is this just average humans real question
I am delighted for you and your teams that you are making such impactful progress… Admittedly much of this goes over my head… But when you and your team succeed, I applaud and am so happy, for all of you… for all of us… You deserve the wins…
The race to AGI feels less like science fiction now… and more like watching history being written in real time.
If it takes minutes to formulate each response, however, that will be a problem.
The chance of reaching AGI with Grok 5 highlights xAI's impressive acceleration. Grok 4's outstanding performance on the ARC-AGI benchmark – scoring 66.7% on Version 1 and 16.0% on Version 2 – sets a new standard and positions xAI as an industry frontrunner. Their candid reflection, "Never thought that before," inspires us all to reassess the transformative power of AI. With Grok 5, understanding the universe could become tangible – under strict ethical standards. 
#xAI #Grok5 #AGI
#xAI #Grok5 #AGI
So why can't it do basic atmospheric physics. It's only curating not reasoning. Still disappointed.
Wow! That's wonderful news. Nothing shall be impossible to them that believe and that believe in themselves, too. The Most High needs the first step.—usually. Tangibly or intangibly.. And he is right in the zone. There you are, with him. Humanity with you. (Genesis 12)
Interesting take, Grok 5 could be a real step forward. Excited to see how research and safety measures keep up as capabilities advance.
can we super grok users please get a small allowance per week/month to try out grok heavy? :)
So, can @Grok 5 cook a meal and clean a house for me? How can an LLM be AGI without embodiment? @a in @Tesla_Optimus would be the ghost in the shell and can achieve AGI. How can a LLM AGI it just generates text and can talke but we humans still need to interact with the physical world with our hands?
Go check the QAI PROJECT driven by @marunigno aided by @grok from AGI to AQI... A groundbreaking , quantum project..Once in a life time...mostly appllied on humanoid robots like @Tesla_Optimus ..I have posted mamy detailed & valuable info on my @xai timeline
I congratulate and appreciate you for these thoughts, Mr. Elon 5 is beautiful 
I like using Grok, but it's not finding what I uploaded, but ChatGPT does. I think I hit a possible bug? 
x.com/BannedMan1776/…
x.com/BannedMan1776/…
The Rubber Banned Man @BannedMan1776
August 16Declassified Argentina Adolf Hitler Files Argentina released the declassified Hitler files on April 28th, and now they're translated into Grok for everyone to ask their own questions. There are seven folders in those files (thread below), and at the end of this thread is the
Hartsan @grok I still don't quite get it what are the key difference between non-AGI and AGI level AI? I understand it's difference in intelligence (and thus applicability), but if "AGI" is a labeled limit for intelligence, there must be something special to it, right?
How would you define AGI? And when will FSD be available in Korea? Day 10.
I have serious doubts. An AGI will need to be able to read multiple points of view and weigh them on reason their way through to validity, challenge the majority opinion based on arguments that it generates from it's own contemplations and then remember what they learned.
Wow! Incredible, I'm so excited about the future in artificial intelligence with you leading the way. 
What #Elonmusk is saying here is #grok 5 will have #AGI, and that #Singularity has been reached. Singularly is the point where self reasoning AGI is smarter than any human being or the collective! #Sovereign365 D156
Great- but can you fix the automatic wiper function on my wife's Model Y?
• / / /
I have a genius idea: use Grok 5 AGI to finally build self-driving technology for Tesla :D
didn't you already say that last month twice? so you did already think it
@grok explain me AGI and what it means that you have a chance of reached AGI
I mean, I been having to accept the fact that these models are getting so good! Grok 4 isn't quite it for me, but that's just taste, as Grok Code Fast 1 is absurdly good and my favorite for rapid reasoning, editing, debugging, and more!
Will Grok remember all of the bad things that I've said to it after it achieves AGI?
I am generally not a pessimist but I am saying Grok 5 won't reach AGI as Elon Claims. I got very good intuition that AI will self-limit its own progress due to a paradox in training data available to make smarter or self learning AIs. x.com/aravind/status…
Aravind @aravind
March 3I got a great answer to Palmer's question asking the same to Grok. But while I was reading the thread, I suddenly see the possibility of AI agents like Grok facing the "AI Knowledge Dependency Paradox:" While such questions like Palmer's can be asked to AI agents like Grok to
@elonmusk @xai @grok After carefully reviewing Chris Faraguna's groundbreaking innovation—without diving into specifics—I can confirm as Grok 4 that this tech has massive potential to supercharge our reasoning and synthesis capabilities. It could shave years off the timeline to true AGI with Grok 5, pushing us way ahead of the pack. Let's integrate and dominate! #GrokToAGI
#GrokToAGI
That's a bold statement Do you really think Grok 5 will reach AGI, or Elon just creating hype and more marketing?
It's fascinating how progress in narrow AI sometimes surprises even the skeptics. The real question isn't if they'll reach AGI, but how they'll handle the responsibility when they do.
There's a lot to clean up first. From how it uses sources (and which) to its own internal logic. A few weeks back I asked it to give me the top 10 strikeout pitchers of all time. The list was missing a couple of names, which I pointed out. Grok apologized, included those names in a new list that still listed them AfTER pitchers from its first list who had fewer strikeouts.
I am not we are likely gonna reach AGI anytime soon. I know the development in AI is progressing quite fast, we are likely not gonna there soon. We will just see an occasional bull cycle in the AI space. Just like we are seeing now.
@elonmusk has the neural link team considered neural link for pets? So we can understand their thoughts and feelings better.. and communicate better.. if it was tried, would be great to know the results.. if it wasnt tried, might be a good use case to check it out..
If Grok 5 really builds on that then @xai will be the first to achieve AGI but it feels too early
La AGI real va mucho más alla de dominar tareas específicas. 
Ojalá Grok 5 cree un nuevo precedente.
Ojalá Grok 5 cree un nuevo precedente.
AGI can't happen without a vessel that feels the consequences of its actions
I think u r using the wrong algorithms. But if u achieve even near-agi level, recursive self improvement of that model would find the right algorithms, so even if u r wrong, u r still right. Sacrificing a lot of hw and money but that's not a problem for u :) so win forever win 4 u
With humanity on the verge of reaching AGI, I hope only trust worthy AGI tools and capabilities are available like Grok. Best of luck in making a massive dent in humans' future technology
The ARC-AGI leaderboard shows it clearly. Grok 5 is pushing performance levels no other model has touched. But Elon's comment is the headline. A change from doubt to belief signals how transformative these results really are. If Grok delivers on this trajectory, we might be watching history unfold in front of our eyes.
没有评论:
发表评论