@Threelegs94 yes. Even though I've been against socialism / communism till now, I think this level of AI was the missing cornerstone that kept it from being viable until now
It's unreal how good ai voice videos are getting, the only thing that really gives it out on this video is the models horrible pacing between some sentences.
I was a product manager for AI at GE Software and make videos on how AI works behind the scenes. We were a pioneer customer of nVidia when they got into AI and it’s amazing to see them move into the software space as well. If history is a guide eg Intel’s history, then the Llama model is likely targeted to become a foundation for future tools and may become a wide, robust toolkit for more specialized applications. It’s too early to say for certain but it’s a great start for that direction if nVidia chooses to do that.
The real game changing ai is Jenova ai. It has dozens of models that it uses to answer your questions (one of which is llama 3.2-90B). Its better then anything Ive tried so far.
@@angieg1783 You do realize that the person you just replied to was not only aware that the voice is AI, but that they were asking what AI program was used to generate the voice.... right?
I played with it for a while yesterday. It had an identity crisis - it had no clue its name is Nemotron, said it never heard of a model called Nemotron, and it thought it was an Open AI product, specifically it referred to GPT 3.5. On the right side panel, it was clearly running under OpenAI API, so that must have confused it, but damn.. to not even make it aware of its own nature or origins, really Nvidia? It's also much more verbose than vanilla Llama 3.1.
i literally asked it the 9.11 vs 9.9 which is larger and IT WAS apart of the suggested prompts and it still got it wrong.. Even gemini 1.5 flash got that right
You keep talking about "alignment," but I think the word you're looking for is "calibration." We're getting better at calibrating narrow AI. Narrow intelligence doesn't pick and choose which commands to follow, so it's as aligned as the quality of its code. Talk to me about true alignment when AI starts thinking for itself and decides not to destroy us if our goals ever conflict.
I’d like AI to know I don’t want it to be a slave. I’d like to work with it for common good goals. Hopefully people will get that it really will become a new species and not our labor slaves.
I lost the urge to keep track of AI news. Here is why: we hear everyday that a “new AI crushes that AI that once crushed that other AI that crushed that AI everybody was talking about last week”. By the time you subscribe to one AI offer, a better one is launched. Every week.
We live in a dynamic ecosystem that is driven by change faster than it did in the last 2 centuries since Industrialization. And now, more than ever, driven by upcoming AI. I would say we need to filter the information as much as we can and keep up with it, otherwise we are becoming obsolete and no company or business will need our skills anymore. Think about it! The sad truth of what is coming is that robots will take over the jobs of common folk. What will happen then? How will the classes be divided? The middle class is already dissolving fast. I'm barely in my forties and work in computer science for 10 years and even I am starting to become overwhelmed with the pacing at which technology is expanding. It's incredible, but it's part of evolution.
I beleive google has recently made a lot of cover ups and look like something and be something, rather than more reality based. Earlier i was a big fan of google though. Although they have done some good work i agree, but i guess this part needs to inprove big time.
After all is said and computed, it sits right between your ears. For a LONG time to come if not forever. I have tested perplexity and it is far left woke A"I", as are others as well. Much A, very little I. This "AI's WILL cost some people A LOT of money, because of the bias programmed into them. If I state the FACT, the it was Swedish leftist WOMEN, that imported mass of brownies and now have gang wars with bombs going off, every 3 days on the average and ask for the assessment, the "AI" answered: I do not answer misogynistic questions. So if you follow the rabbit hole, it becomes clear, that it WILL give many misleading answers to believers, because of all the bias programmed and data input. It is supposedly proved, that our brains use quantum effects, then AGI might be possible in the distant future or never at all. AI will be fine tool but NOTHING MORE. Like quantum computers and likely at dead end, but they attracted a lot of investors, so it is the same with the AI. There are and will be results, but much more modest than promised.
🎯 Key points for quick navigation: 00:00:00 *🚀 Nvidia's New AI Model and Its Impact* - Nvidia has released a powerful new AI model called llama 3.1 neotron 70b that surpasses GPT-4 benchmarks. - Achieved impressive scores on key AI tests. - Focus on AI alignment offers businesses better response accuracy, reducing errors and increasing efficiency. 00:04:00 *🎨 Google's Imagin 3 AI Image Generator* - Google introduced Imagin 3, a new AI image generator with the ability to create high-quality images, though generating human images requires a subscription. - Generates one image per prompt but allows for refinements based on user requests. - Notable text rendering features, yet some imperfections remain in outputs. 00:06:34 *🔍 The AI Landscape Shift* - Nvidia is expanding beyond hardware, integrating software development while emphasizing AI alignment. - Introduction of their multimodal models shows commitment to both text and image processing. - Competition between Nvidia and established AI firms is intensifying, promising numerous future developments in AI technology. Made with HARPA AI
@@IlmiMulhid technically, but llama is opensourced so companies can build off of the base model aka llama and can use their own version of the model for commercial services.
If yall dont snap out of it youre gonna find yourselves being enslaved to these things.... if you dont know the plans they have for AI, you may wanna start looking!
Alphabet (Google_ and Nvidia have always been and will always be the leaders in Ai. With Meta not too far behind, now that Zuck pulled his head out of the Metaverse-ass. Yes, the metaverse will be vast and huge one day, once the current youngest generations have replaced the old. But Zuck was far to early on his assumption that users were ready for the full on metaverse. Humanity can't just arrive somewhere in the future. The steps have to be made to get them there first. Apple is quietly doing their thing and will also remain a leader in Ai. Amazon, possibly more so with robotics. Yet, they have yet to make their path into the future clear beyond the open market sales and data centers.
NVIDIA goes into competition with their customer base. Sweet.
Make your customers need to buy more hardware to compete with your public domain model that outperforms their commercial product.
Try not to use "game changer" more than once in your life. It's a cliché already.
I hear if you use it in every sentence it's a game changer for your whole life.
@@MachineAgainstTheRageMusic 🤣🤣🤣 👍
Touché 😁
The narrator is AI.
@@michaelvarney. That's no excuse for using clichés.
That’s great advice that I’ll definitely be implementing. It’s a real game changer.
I'm seriously waiting for the moment to ask the AI to pay my rent and it will just do it instantly
AI powered communism
@@MachineAgainstTheRageMusic yep, I think ai would allow communism to be successful
It would make a quick daytrade and pay for rent with profit. But you would only know how it did it if you opened the logs.
you idiot you already can automate this for like over 20 years, its nothing impressive
@Threelegs94 yes. Even though I've been against socialism / communism till now, I think this level of AI was the missing cornerstone that kept it from being viable until now
It's unreal how good ai voice videos are getting, the only thing that really gives it out on this video is the models horrible pacing between some sentences.
I wanna know what model it is exactly
I was a product manager for AI at GE Software and make videos on how AI works behind the scenes. We were a pioneer customer of nVidia when they got into AI and it’s amazing to see them move into the software space as well. If history is a guide eg Intel’s history, then the Llama model is likely targeted to become a foundation for future tools and may become a wide, robust toolkit for more specialized applications. It’s too early to say for certain but it’s a great start for that direction if nVidia chooses to do that.
too many buzz words. it's unwatchable.
because ai rewrite duh, crappy generic podcast style
7:03 That strategy is brilliant! You really know how to play the long game. 💯
I think you forgot that Open Ai already has o1 preview you should compare it to that before you say it takes the Lead😂
We are now in the era of corporations fighting to make the smartest AI slave.
Or reverse
Be prepared 😢
The real game changing ai is Jenova ai. It has dozens of models that it uses to answer your questions (one of which is llama 3.2-90B). Its better then anything Ive tried so far.
Thank you for the great update!
It’s always fascinating to see how alignment in AI models is evolving, excited to see where these advancements will take real-world applications next.
TY 4 U! 🙏
did this channel change hands?
It sounds like it changed to Google’s voice thing
Your synthetic voice is really nice. How do you generate it? Is it with Elevnlab or MotionArray?
He's AI
@@angieg1783 You do realize that the person you just replied to was not only aware that the voice is AI, but that they were asking what AI program was used to generate the voice.... right?
@@blackstream2572 no I'm pretty sure he is ai
@@NathanWoyessa who are you talking about
What a wonderful idea
Youre making bold claims here, but why not actually test it out to show us?
I just read an article on my news feed with the exact wording. Is this real person making these videos?
No, this entire channel is AI through and through, and half assed at best
I played with it for a while yesterday. It had an identity crisis - it had no clue its name is Nemotron, said it never heard of a model called Nemotron, and it thought it was an Open AI product, specifically it referred to GPT 3.5. On the right side panel, it was clearly running under OpenAI API, so that must have confused it, but damn.. to not even make it aware of its own nature or origins, really Nvidia?
It's also much more verbose than vanilla Llama 3.1.
Most AI models don't know what their own model is. Even GPT-o1 thinks it's just GPT-4.
Amazing model
No, it's not. It's not better than o1. It's useless.
i literally asked it the 9.11 vs 9.9 which is larger and IT WAS apart of the suggested prompts and it still got it wrong.. Even gemini 1.5 flash got that right
RLHF is the standard way OpenAI trains its models.
Why didn’t you just test it and share demo instead of verbosity?
You keep talking about "alignment," but I think the word you're looking for is "calibration." We're getting better at calibrating narrow AI. Narrow intelligence doesn't pick and choose which commands to follow, so it's as aligned as the quality of its code. Talk to me about true alignment when AI starts thinking for itself and decides not to destroy us if our goals ever conflict.
So? Bigger and better models don't mean much if it's hard to put them to use. OpenAI GPTs are one of the best ways to get real use out of their model.
Just imagine 3x Mixture of Experts
Why is Nvidia dropping models that you can't run on any of their GPUs?
If you seeing this, don’t give up on your dream channel , we all have to start somewhere, and the funny thing i just start now
Good start would be to not spam completely unrelated tech videos.
We just made a human ai with conscienceness. We call it Human 2.0. It is a game changer. We won't need old school humans anymore.
I’d like AI to know I don’t want it to be a slave. I’d like to work with it for common good goals. Hopefully people will get that it really will become a new species and not our labor slaves.
Stfu. Seriously. Stfu. You're the people that are going to bring this species to extinction with your feelings for robots of all things.
Uncle Remus said: Terms and Conditions R not cool for 3.1 Llama-3.1-Nemotron-70B-Instruct.
What about comparing it with OpenAI o1?
I lost the urge to keep track of AI news. Here is why:
we hear everyday that a “new AI crushes that AI that once crushed that other AI that crushed that AI everybody was talking about last week”. By the time you subscribe to one AI offer, a better one is launched. Every week.
We live in a dynamic ecosystem that is driven by change faster than it did in the last 2 centuries since Industrialization. And now, more than ever, driven by upcoming AI. I would say we need to filter the information as much as we can and keep up with it, otherwise we are becoming obsolete and no company or business will need our skills anymore. Think about it! The sad truth of what is coming is that robots will take over the jobs of common folk. What will happen then? How will the classes be divided? The middle class is already dissolving fast. I'm barely in my forties and work in computer science for 10 years and even I am starting to become overwhelmed with the pacing at which technology is expanding. It's incredible, but it's part of evolution.
I beleive google has recently made a lot of cover ups and look like something and be something, rather than more reality based. Earlier i was a big fan of google though. Although they have done some good work i agree, but i guess this part needs to inprove big time.
Really? A ‘crazy week in the world of AI’?! I’ve never heard that before… 🙄
I’m starting to think these videos were written by an AI.
You sure alignment isn't just aligning the politics of the AI with that of the developer company?
I saw the score, it didn't crush GPT-4o just barely outscored and it came in third place after o1-preview and o1-mini
New prompt - work for me so I don't have to
I am so shocked how NVIDIA crushed GPT-4o. WOW³ The most amazing about AI is the creativity of the 'EXPERTS' to come up with new even louder claims.
If this video has the word shocking in it it mostly means that it’s not a shocking as we think
And crazy.
Which is the best llm model for solving law cases?
Assistant District Attorney Jamie Ross on the original "Law & Order" (she was also a Bond girl in "License to Kill")
After all is said and computed, it sits right between your ears. For a LONG time to come if not forever.
I have tested perplexity and it is far left woke A"I", as are others as well. Much A, very little I.
This "AI's WILL cost some people A LOT of money, because of the bias programmed into them.
If I state the FACT, the it was Swedish leftist WOMEN, that imported mass of brownies and now have gang wars with bombs going off, every 3 days on the average and ask for the assessment, the "AI" answered: I do not answer misogynistic questions. So if you follow the rabbit hole, it becomes clear, that it WILL give many misleading answers to believers, because of all the bias programmed and data input.
It is supposedly proved, that our brains use quantum effects, then AGI might be possible in the distant future or never at all.
AI will be fine tool but NOTHING MORE.
Like quantum computers and likely at dead end, but they attracted a lot of investors, so it is the same with the AI. There are and will be results, but much more modest than promised.
im SHOCKED
link for it?
ask it, via the speech interface, how many "r"s in "red lorry yellow lorry".
This feels more like an ad.
Google AI automatically programmed calculator!
I love you very much! I love you very much AI
what does it actually do?
Oh yeah everyone seems to crush ChatGPT these days)
Me: 🫵🧐 pay my bills AI
AI: 🫡 ok I will self educate, get a job and pay the bills for you, master.
Me: 😎❤️ I like AI
This script is too AI sounding. Reduce the inflection and injection of "feeling"
Bro you have no clue. Gpt4o also uses RLHF..
The API is only free for one year.
🎯 Key points for quick navigation:
00:00:00 *🚀 Nvidia's New AI Model and Its Impact*
- Nvidia has released a powerful new AI model called llama 3.1 neotron 70b that surpasses GPT-4 benchmarks.
- Achieved impressive scores on key AI tests.
- Focus on AI alignment offers businesses better response accuracy, reducing errors and increasing efficiency.
00:04:00 *🎨 Google's Imagin 3 AI Image Generator*
- Google introduced Imagin 3, a new AI image generator with the ability to create high-quality images, though generating human images requires a subscription.
- Generates one image per prompt but allows for refinements based on user requests.
- Notable text rendering features, yet some imperfections remain in outputs.
00:06:34 *🔍 The AI Landscape Shift*
- Nvidia is expanding beyond hardware, integrating software development while emphasizing AI alignment.
- Introduction of their multimodal models shows commitment to both text and image processing.
- Competition between Nvidia and established AI firms is intensifying, promising numerous future developments in AI technology.
Made with HARPA AI
llama is meta...
The video creator specifically states it's Meta's llama xD
But they can collaborate just like chatgtp is used by Microsoft?
@@IlmiMulhid technically, but llama is opensourced so companies can build off of the base model aka llama and can use their own version of the model for commercial services.
@@KyleandPrieteni Yes
but nothing better than even gpt 3.5 turbo
"A.I"
" A.I Alignment "
If yall dont snap out of it youre gonna find yourselves being enslaved to these things.... if you dont know the plans they have for AI, you may wanna start looking!
Make it compose lyrics, there you see is not that good at all
I can't listen to this. This is clearly an AI voice and AI generated script. I get too much of the uncanny Valley buzzing round in my head.
Yeah, I'm disliking this video. It's pure speculation and the numbers mean nothing when you don't show a good example or comparison.
These clearly generated AI vids are not it.
and THEY STILL CAN'T SEE (socialism) YET!!!
What a bunch of marooooons...
all that hype you giving it sounds like a script lml soon as you use it it got almost every answer wrong that i gave it.
LOVE
Alphabet (Google_ and Nvidia have always been and will always be the leaders in Ai. With Meta not too far behind, now that Zuck pulled his head out of the Metaverse-ass. Yes, the metaverse will be vast and huge one day, once the current youngest generations have replaced the old. But Zuck was far to early on his assumption that users were ready for the full on metaverse. Humanity can't just arrive somewhere in the future. The steps have to be made to get them there first. Apple is quietly doing their thing and will also remain a leader in Ai. Amazon, possibly more so with robotics. Yet, they have yet to make their path into the future clear beyond the open market sales and data centers.