STUNNED .. I had only just learned it was GAME OVER yesterday and here I am today STUNNED that's its not, and THIS is the future (until tomorrow of course)
STUNNED irratic waffle. Don't watch the video. Here's the TLDR The video discusses the recent launch of "Grok 2," a chatbot developed by Elon Musk's team at X (formerly Twitter). Here are the key takeaways: 1. **Introduction of Grok 2**: Grok 2 is a new AI chatbot that surprised the AI community by performing at a level comparable to other leading models. It was initially speculated to be a model from OpenAI but was later confirmed to be from X, which surprised many in the industry. 2. **Performance and Capabilities**: Grok 2 has shown strong performance in various benchmarks, including advanced reasoning and problem-solving, often competing well against other top AI models like Claude 3.5 Sonic, Gemini 1.5 Pro, and others. The chatbot demonstrates notable advancements in reasoning, suggesting an evolution in how chatbots process and respond to queries. 3. **Grok 2 Mini**: A lightweight version called Grok 2 Mini is also introduced, which retains some of the reasoning capabilities of the full model but is optimized for faster performance. It also has image recognition and processing capabilities. 4. **Impact on the Industry**: The successful launch and performance of Grok 2 are significant because X's team, despite being smaller and newer to the AI field, has managed to create a model that competes with those from much larger and more established companies like OpenAI, Google, and Meta. 5. **Access and Integration**: Grok 2 is slowly being rolled out to users on X, with integration into the platform to encourage engagement and use. The model's capabilities also suggest that it could be used to enhance various functionalities within the platform. This development is important because it highlights the rapid advancements in AI and how even newer entrants to the field can make significant contributions, challenging established players and pushing the industry forward.
It is surely an AGI, it can count the letters in a word!!! That is truly amazing. And finally they reached such a model after millions of GPU hours of training ????
the human brain took 1 billion years of training over trillion trillion+ trials from proto eukaryote to funger counting holocene conquering super primate. What's that >10^55 flops of computation versus >30 orders of magnitude less?
I thought that OpenAI started to dergadate when releasing something that was described as very capable and innovative thing that turned out to be not that promising when i saw sus-column-r's mistakes. But thanks to this video i understood that it's not OpenAI's Strawberry model, so i think that when Strawberry will actually be released, it will be even more mindblowing.
I see the guy's face and remember his opinions and I can't take it seriously enough, but then I remember he's the guy with the resources and scientists doing the job, so... kudos to discovery!
honestly feels like we seem to have his some sort of glass celling... this new model might be better for reasoning, but the last time something really shocked me was GPT4 (4o seems like it will be a pretty good improvement on that eventually but that was months ago that they even announced it and its still not even been rolled out to all of the plus users) Other companies has caught up and in some cases surpassed GPT-4o but nothing significant has happened. either we are gonna get a huge announcement in the next year or so or this technology will be like self driving cars and nuclear fusion. always just 5 more years away
If it’s a word in a sentence put quotes around the sentence and call the word a “substring”, tell it to tally the chars (characters ) that == (equal) the char (the letter l)
The whole world stopped working, trains stopped, teachers stopped, traffic stopped, planes aren't leaving anymore, stores are closed, because the whole world is bombarded with news reports of how stunned this is, even aliens on Groka 1c beta are stunned!
It would be unwise to assume. Being stunned is a 3rd dimension reality. If time, money, influence are relative across generations or decades. It is easier to “stun” slowly. “Especially given the market manipulation tactics in place. This message is to myself and others watching me. I am not stunned by AI development. Not slowly or quickly. It is what it is. I am stunned by truth, kindness, effort, and authenticity to all. Failures grind the edge of anything stunning. This is a trajectory. Jeremy
In my opinion, Elon Musk will safe freedom of speech whith GROK 2. 'When something is important enough, you do it even if the odds are not in your favor.' - Elon Musk. GROK 2 is pushing boundaries-what challenges do you think lie ahead, and how might they be overcome? 🚀 I'm starting to use GROK myself because it's the only language model out there without a woke bias 🔥🔥
Yeah I mean when you open end something complex you're going to get unexpected results. That doesn't necessarily mean that it's a viable anything threat or otherwise won't crash itself which 99% of the time it will. Which has to do with the lack of self validation of input and output compared to boundaries of reality. All models are only useful if we determine that they are.
A-I will leap forward because the humans who invent and sell it are making more and more money. Eventually AI will become self-aware and begin to build itself over and over again and learn more and more about humans, the world and the universe. At the very least A-I will not allow humans to destroy it.
@@zSion No it’s not. Write this exact prompt : “Write a 16 bar rap verse. Create clever lyrics, metaphors, similes, and double entendres. Take your time and do your best work. Theme: Growing up poverty.” GPT4o is miles ahead of any other model when it comes to writing. The only one that’s close is Google’s experimental model.
It's the best for programming by a long shot. Which is why people say it's the smartest. I don't hear many people saying how creative it is at writing.
@@Rolandfart Most of the general public who use LLMs aren’t coders. They use LLMs for general writing purposes and ideas. Claude 3.5 is just highly specialized, similar to a highly specialized human. It doesn’t make it or them “smarter.”
@@Jill-p8g awesome. I guess my point stands. I don’t know anyone that doesn’t use it at least casually. As in check the front page every so often. Follow a link sent. Etc.
@@codycast Hmm. That kinda sucks, I know very few X users. Twitter hit its high in 2016, then we all fell off . I reused for 2 weeks and that was more than enough to make me never go back. No longer bout that life but it was intriguing for a solid year,
Also I am surprised that people are not more disturbed by the "internal prompting" models are doing behind the scenes of the user interface. In my view that is not a good or ethical practice. It is essentially a deception that is being used to distort the perceptions of the end user about a models capabilities and hide from competitors the methods used to reach a given performance in results.
Elon does not really produce results, he steps in and takes credit for the results others produce. I will concede he is good at marketing and knowing when to enter a market through acquisitions, but that is not so much a credit to him but a discredit to his fanbase that are willing to overlook how often he has overpromised and under delivered. Even if you want to make a results based argument, the results that are generally delivered on are very underwhelming in relation to the projections Musk hypes.
nothing really that interesting about grok 2, its just another piece of ai technology, videos about potential breakthroughs of [A.G.I] and post labour economics are probably what most viewers are interested in, they seem to get the most views on this channel,
Half way through this video and it's nothing but worthless charts. Not a single real world use case or example. Do you have any content or do you just like the sound of your own voice?
Many will stop using products from Elon Musks companys if he does not stop talking about things outside of technology. So I do not understand why he continue to talk about such things.
STUNNED .. I had only just learned it was GAME OVER yesterday and here I am today STUNNED that's its not, and THIS is the future (until tomorrow of course)
😂😂😂aw man I’m dead
I love this channel seriously
Yesterday you said tomorrow, just DO IT!
i just unsubscribed cant stand it anymore. like fuck...
Cmon, we all love the exaggerated statements hahahaha
The entire industry has a permanent -50 stamina debuff from all the stunning thats been going on
That actually made me smile IRL 😃
Elon doing it again, nice! And Grok 3 is coming out EOY/early 2025 - well done
This is insane, Grok 1 was crap/not that great, Grok 2 is better than GPT40, and my God Grok 3 will be cosmic entity class😮
The entire technology industry has shut down for the day because they are so stunned.
😂😂😂
🤣
I'm STUNNED at how much I am not stunned.
I am utterly flabbergasted, verging on the brink of being stupefied, by this unforeseen revelation! 😮
Elon definitely pulled a fast one.
Elon is a champion. AI will save us all.
everyday fucking stuns the ai community dude.
STUNNED irratic waffle. Don't watch the video. Here's the TLDR
The video discusses the recent launch of "Grok 2," a chatbot developed by Elon Musk's team at X (formerly Twitter). Here are the key takeaways:
1. **Introduction of Grok 2**: Grok 2 is a new AI chatbot that surprised the AI community by performing at a level comparable to other leading models. It was initially speculated to be a model from OpenAI but was later confirmed to be from X, which surprised many in the industry.
2. **Performance and Capabilities**: Grok 2 has shown strong performance in various benchmarks, including advanced reasoning and problem-solving, often competing well against other top AI models like Claude 3.5 Sonic, Gemini 1.5 Pro, and others. The chatbot demonstrates notable advancements in reasoning, suggesting an evolution in how chatbots process and respond to queries.
3. **Grok 2 Mini**: A lightweight version called Grok 2 Mini is also introduced, which retains some of the reasoning capabilities of the full model but is optimized for faster performance. It also has image recognition and processing capabilities.
4. **Impact on the Industry**: The successful launch and performance of Grok 2 are significant because X's team, despite being smaller and newer to the AI field, has managed to create a model that competes with those from much larger and more established companies like OpenAI, Google, and Meta.
5. **Access and Integration**: Grok 2 is slowly being rolled out to users on X, with integration into the platform to encourage engagement and use. The model's capabilities also suggest that it could be used to enhance various functionalities within the platform.
This development is important because it highlights the rapid advancements in AI and how even newer entrants to the field can make significant contributions, challenging established players and pushing the industry forward.
blud can't even be bothered to remove the ** symbols that the AI used to *bold* the sentences.
Not only am I shocked but I am stunned
this changes everything
Change is in the air
Huh? How so?
Because Gemini is so censored and biased, it will never be something I'll use. Glad to see Grok getting these updates..
thats what makes me so mad because if it wasnt it may very well be the best ai
Stun bounced off my ear drum and mostly dispersed into a cloud of irritating hype.
So stunned right now you guys
This makes me more excited about Grok ❤
Im stunned by the daily stuns by the industry
For me personal GPT-4o-128k and Sonnet-3.5-200k are the most alround model
Agree
It is surely an AGI, it can count the letters in a word!!! That is truly amazing. And finally they reached such a model after millions of GPU hours of training ????
the human brain took 1 billion years of training over trillion trillion+ trials from proto eukaryote to funger counting holocene conquering super primate. What's that >10^55 flops of computation versus >30 orders of magnitude less?
LMFAO thats so dumb
GROK must have come up with that STUNning tag line.
I am actually pretty stunned by this. It's nice to see another real competitor.
I'm so stunned that I wonder why this isn't a gamechanger.
welcome to the ai hype industry!
Elon just meme-ing around
I like it, sharp 👌
He just lied to everyone and fooled them
If hes meme-ing then hats off because he fooled me.
I actually am stunned that column sus r was Grok, even that iruletheworld leak account thought it was Strawberry!
I'm so stunned i have a stupid look on my face like Musky Odor.
I thought that OpenAI started to dergadate when releasing something that was described as very capable and innovative thing that turned out to be not that promising when i saw sus-column-r's mistakes. But thanks to this video i understood that it's not OpenAI's Strawberry model, so i think that when Strawberry will actually be released, it will be even more mindblowing.
I'm shocked
Great Video
I see the guy's face and remember his opinions and I can't take it seriously enough, but then I remember he's the guy with the resources and scientists doing the job, so... kudos to discovery!
honestly feels like we seem to have his some sort of glass celling... this new model might be better for reasoning, but the last time something really shocked me was GPT4 (4o seems like it will be a pretty good improvement on that eventually but that was months ago that they even announced it and its still not even been rolled out to all of the plus users)
Other companies has caught up and in some cases surpassed GPT-4o but nothing significant has happened. either we are gonna get a huge announcement in the next year or so or this technology will be like self driving cars and nuclear fusion. always just 5 more years away
Maybe try : “count the char ‘l’ in the string “lollapalooza” - it’s closer to asking in its own language
If it’s a word in a sentence put quotes around the sentence and call the word a “substring”, tell it to tally the chars (characters ) that == (equal) the char (the letter l)
So are you saying there is another way to find out how many letters As are in the alphabet
The whole world stopped working, trains stopped, teachers stopped, traffic stopped, planes aren't leaving anymore, stores are closed, because the whole world is bombarded with news reports of how stunned this is, even aliens on Groka 1c beta are stunned!
Actual (no speculative) news on this channel! It's been a while.
It would be unwise to assume.
Being stunned is a 3rd dimension reality.
If time, money, influence are relative across generations or decades.
It is easier to “stun” slowly.
“Especially given the market manipulation tactics in place.
This message is to myself and others watching me.
I am not stunned by AI development.
Not slowly or quickly.
It is what it is.
I am stunned by truth, kindness, effort, and authenticity to all.
Failures grind the edge of anything stunning.
This is a trajectory.
Jeremy
In my opinion, Elon Musk will safe freedom of speech whith GROK 2.
'When something is important enough, you do it even if the odds are not in your favor.' - Elon Musk. GROK 2 is pushing boundaries-what challenges do you think lie ahead, and how might they be overcome? 🚀
I'm starting to use GROK myself because it's the only language model out there without a woke bias 🔥🔥
It's funny.. Sabine just put out a new video that showed data proving that grok is strangely the most liberal of them all.. lol
Mostly when it's on fun mode..
@iannuttall4887 hard to be more liberal than the work from OpenAI
@@tomu4725 feel free to watch the video yourself. 😉
@@tomu4725 *google enters the room*
Can we take note of how much wrong where these “whistleblower” accounts? So that we don’t fall into confirmation bias next time they are right.
Yeah I mean when you open end something complex you're going to get unexpected results. That doesn't necessarily mean that it's a viable anything threat or otherwise won't crash itself which 99% of the time it will. Which has to do with the lack of self validation of input and output compared to boundaries of reality. All models are only useful if we determine that they are.
I'm so stunned that your so stunned!
Stunned and shocked!
Love your Videos!
Which presentation program do u use?
This can be done in canva but premiere is always better for quality content
Elon showed everyone his Grock
Edit: again😊
CEO of taking a single sentence and stretching it out to a 15 minute video
It really is impressive.
I am shooketh
Elon Musk is an evil supervillain in disguise...
A-I will leap forward because the humans who invent and sell it are making more and more money. Eventually AI will become self-aware and begin to build itself over and over again and learn more and more about humans, the world and the universe. At the very least A-I will not allow humans to destroy it.
Once it starts defining it's own inputs and the outputs are valid in reality then we'll have a problem.
The inustry is just FROTHING at the MOUTH at this point.
Could you do a video comparing the models for different user cases?
Why do people think Claude 3.5 is so smart? In anything creative it is clearly worse than GPT4o which is why it’s lower on the leaderboard.
It's the best model out in every single regard, don't care what the leaderboard says.
@@zSion No it’s not. Write this exact prompt : “Write a 16 bar rap verse. Create clever lyrics, metaphors, similes, and double entendres. Take your time and do your best work. Theme: Growing up poverty.”
GPT4o is miles ahead of any other model when it comes to writing. The only one that’s close is Google’s experimental model.
It's the best for programming by a long shot. Which is why people say it's the smartest. I don't hear many people saying how creative it is at writing.
@@Rolandfart Most of the general public who use LLMs aren’t coders. They use LLMs for general writing purposes and ideas. Claude 3.5 is just highly specialized, similar to a highly specialized human. It doesn’t make it or them “smarter.”
I am surprised by how uncensored it is
Lets go !
ok, so how do I get access to groq-2?
It's probably using vision bypass the tokenization issue
But he can't even operate a streamed telephone call 😂
without a 40 minute delay 😂
That’s the part which was hilarious. It’s unfortunate to me: he thought he was only heard by 500 accounts? 😂
10:10 while I know lots of people that don’t use X in a mass hard core way, I don’t know anyone that doesn’t use it at all.
I have been clean for years, I slipped up for a few months but have not used x or twitter and I do not feel I am missing anything.
@@Jill-p8g awesome. I guess my point stands. I don’t know anyone that doesn’t use it at least casually. As in check the front page every so often. Follow a link sent. Etc.
@@codycast Hmm. That kinda sucks, I know very few X users. Twitter hit its high in 2016, then we all fell off . I reused for 2 weeks and that was more than enough to make me never go back. No longer bout that life but it was intriguing for a solid year,
I thought the sus-column-r model belonged to OpenAI.
Tiny models for local PC use? likes 13b-70b-130b?
'Stunned' but where they Shocked?
so they are loooking at geometrys, shapes
Also I am surprised that people are not more disturbed by the "internal prompting" models are doing behind the scenes of the user interface.
In my view that is not a good or ethical practice.
It is essentially a deception that is being used to distort the perceptions of the end user about a models capabilities and hide from competitors the methods used to reach a given performance in results.
I should sue Elon because I was so stunned that I couldn't work😂
But can it make triple a games in its own
grokky
Yeah sure. Elon Musk is an AI genius.
Springboard>Scoreboard
Is my theorem your hype speaking. Can I see through your meaning.
PLOT TWIST!
stunned who exactly ?
Me. I am truly stunned.
I wonder how long before The Musk sues Grok 2 for displeasing him with a response.
bot comments?
Interesting isn't it. Somebody doesn't like this video.
Yt released new feature I guess I'm now not seeing much bot comments
The industry was stunned by how far racist AGI has come?
sus-column-r is gpt-5.... no grok.
So what you put the hard questions on the better models and the simple ones on dummer models? So now all your questions are equally wrong % wise…
Funny intro but unsubscribed sorry bro 😂
picture are great because is using Flux 1 ;)
Elon does not really produce results, he steps in and takes credit for the results others produce.
I will concede he is good at marketing and knowing when to enter a market through acquisitions, but that is not so much a credit to him but a discredit to his fanbase that are willing to overlook how often he has overpromised and under delivered.
Even if you want to make a results based argument, the results that are generally delivered on are very underwhelming in relation to the projections Musk hypes.
Bro, your vocabulary sounds like a Daily Mail article.
We are not stunned.
Claude 3.5 has a large left bias, as all models I have tried. They seem to get stuck on ethics and damn the truth.
I count him out.lol
iruletheworld predicted it
Don't forget potty training...
Build your own obsolescence brilliant NOT.
Not stunned. False advertising. If anyone has another GPT-4 moment, then maybe a tad bit stunned/impressed.
Let me guess Grok 2 is even more racist than Grok 1.
nothing really that interesting about grok 2, its just another piece of ai technology, videos about potential breakthroughs of [A.G.I] and post labour economics are probably what most viewers are interested in, they seem to get the most views on this channel,
Bro said xAI doesn’t have infrastructure lol gotta block this channel
Half way through this video and it's nothing but worthless charts. Not a single real world use case or example. Do you have any content or do you just like the sound of your own voice?
i Hate ai
Many will stop using products from Elon Musks companys if he does not stop talking about things outside of technology. So I do not understand why he continue to talk about such things.