@@krellin exactly this. I saw somewhere meta employees saying the same thing. He honestly comes across as a genuinely nice and cool guy if you actually watch his long form interviews and don’t form your opinions based on tabloids.
@@Oliver-eg7vi people just wanted to hate him so they found anything and stuck to it, the whole lizard thing and all the bs... typical bullying except they think they are punching up
I love the local models concept. As a medical trainee, If we could get LLAMA within HIPPA compliant EMRs like epic and cprs it would literally free up half my day worth of notewriting and save lives by giving more time to actually do medicine. Hope those partnerships are in the works
Yes! The AI community is very lucky to have Meta/Zuck push the open source space forward so we can all have state-of-the-art private, localized models for use cases just like this.
By "YouSum Live" 00:00:00 Release of 405 billion parameter model 00:00:05 Llama 3.1 is the most sophisticated open source model 00:01:23 Llama allows community to create custom models 00:02:06 Open source AI could become industry standard 00:03:22 Open source AI leads on cost and customizability 00:04:03 Developers can fine-tune models for specific needs 00:07:34 Meta aims to build a supportive ecosystem 00:10:24 Open source ensures broader access to AI technology 00:11:04 Open source AI promotes safety and scrutiny 00:12:14 Open source can mitigate intentional and unintentional harms 00:17:09 Open source fosters innovation and economic growth 00:25:20 Llama 4 expected to advance AI capabilities 00:27:39 Meta AI aims to be the most used assistant 00:29:03 Every business will have its own AI agent 00:31:10 Open source approach benefits diverse creators and businesses 00:32:05 Skepticism around AI may diminish over time 00:35:02 Open source can create a sustainable political economy By "YouSum Live"
Introduction and AI Announcement - 00:00:00 Meta's LLaMA 3.1 Release - 00:00:59 Comparison with Linux and Open Source Development - 00:02:02 Real World Use Cases and Custom Models - 00:04:01 Partnerships and Ecosystem Building - 00:06:27 Societal Implications of Open Source AI - 00:10:14 Safety and Security in Open Source AI - 00:11:19 Balancing Power and Security Concerns - 00:13:15 Accelerating Innovation and Economic Growth - 00:17:05 Calling Out Apple's Closed Approach - 00:20:04 Restoring the Open Ecosystem - 00:24:16 Future of AI and AGI Vision - 00:27:00 Skepticism and Long-term Commitment to AI - 00:32:13 Implications for Livelihoods and Sustainable Economy - 00:34:15 Closing Remarks and Gratitude - 00:35:13
@@chrishayes5755 It's 405b not 450b, and what you said is not how it works dumbass, usually these models are distilled into smaller partitions capable of running on consumer hardware for general use, by larger labs and companies with compute. Meanwhile larger labs, companies will utilize the full model to provide their own services via their servers for much cheaper in comparison to closed source models, and so on.
@@rowanchI'm really glad to see that he actually knows what he is talking about, he sounds like he is personally invested beyond just profits. I guess he was serious about AI
and great point about windows/apple hardware/software: linux software flourished by being run on commodity hardware that was ubiquitous as a result of windows popularity, though not coupled to its OS inherently
I’m selling all my fancy expensive state of the art gaming items and saving up for future! VR, AI, and robot land here I come. PS Zuck is a fucking beast 🔥 he is on our side these days
Fabric SUMMARY: Mark Zuckerberg discusses Meta's release of the open-source AI model LLaMa 3.1, the importance of open-source AI, and Meta's vision for the future of AI. IDEAS: - Open-source AI can become the industry standard, following Linux's path. - LLaMa 3.1 enables customization, distillation, and fine-tuning of AI models. - Meta envisions many personalized AI agents for businesses and creators. - Open-source AI accelerates innovation by enabling more entities to build. - Meta aims to restore open ecosystems as industry leaders. - Open-source mitigates unintentional AI harms through scrutiny and transparency. - Closed ecosystems incentivize limiting capabilities to box out competitors. - Meta partners to build an open-source AI ecosystem and services. - Open-source AI can help developing countries access leading technology. - Meta plans for multiple future versions of LLaMa models. INSIGHTS: - Open-source enables decentralized AI development across startups, individuals, and nations. - Customizable open models allow proliferation tailored to diverse use cases. - Transparency and collaboration make open-source AI safer than closed development. - Meta aims to democratize AI capabilities through open and personalized models. - Economic incentives drive closed ecosystems to restrict innovation and competition. - Open ecosystems foster more sustainable political economies around emerging technologies. - Meta pursues open AI to avoid being boxed by competitors. - Perpetual open-source leadership provides national security advantages over adversaries. - Open models lift all economies by enabling global entrepreneurship. - Personalized AI agents will become essential for businesses and creators. QUOTES: - "I think this moment with llama 3.1 is kind of like that inflection point where um I think llama has the the opportunity to become the open- source AI standard for open source to become the standard the industry standard for for AI." - "I think that this is a pretty big moment for open source AI um You know I've been reflecting on this and I I kind of think it's you know I thought for a while that open source AI was going to become the industry standard." - "I think that open source is a really important ingredient to having a positive AI future and there are all these awesome things that AI is going to bring." - "It's a little bit Soul crushing when you go build features that are what you believe is good for for your community and then you're told that you can't ship them because some company wants to put you in a box so that they they can better compete with you." - "I want to restore the industry to the state where the open ecosystem is is actually the one that is leading." HABITS: - Meta plans compute clusters and data trajectories years in advance. - Meta aims to make its AI assistant the most used globally. - Meta focuses on enabling creators and businesses to create AI agents. - Meta partners with companies to build open-source AI ecosystem services. - Meta runs research experiments to optimize future LLaMa model architectures. FACTS: - LLaMa 3.1 is a 405 billion parameter open-source AI model. - LLaMa 3.1 is competitive with and outperforms some leading closed models. - Open-source Linux overtook proprietary Unix due to customizability and security. - Mobile development followed a closed ecosystem model benefiting Apple financially. - Over 200 million people on Meta's platforms consider themselves creators. REFERENCES: - Linux - Apple's closed mobile ecosystem - Meta AI assistant - AWS, Databricks (partners for open-source AI services) - Grok AI (low-latency AI inference) - Dell, Scale.ai, Deoe, Accenture (enterprise AI deployment partners) ONE-SENTENCE TAKEAWAY: Open-source AI enables decentralized innovation, empowering businesses, creators, and nations to build personalized, leading AI capabilities sustainably. RECOMMENDATIONS: - Invest in building open-source AI ecosystems and partner services. - Enable customization and personalization of AI models for diverse use cases. - Collaborate transparently to identify and mitigate potential AI safety issues. - Develop tools for businesses and creators to create custom AI agents. - Pursue open development to foster sustainable economic and political landscapes. - Prioritize compute, data, and architectural research for future AI advances. - Educate developers on open-source AI benefits and best practices. - Integrate open AI capabilities into customer-facing products and services. - Advocate for open ecosystems as industry standards over closed models. - Partner with governments on national security implications of open AI.
Follow you on Twitter, and at first thought this was an imposter account because I couldn't believe the amount of subs lol. I expect that to change really quickly. Love your work, and love this interview. The Zuck seems like one of the big tech bros who has the greater good in mind. This is a huge milestone for open source competing against closedAI.
1. 오프닝 - 00:00 2. 메타의 AI 발표 개요 - 00:39 3. Llama 3.1 모델 출시 - 00:53 ㄴ 4050억 매개변수 모델 - 01:00 ㄴ 70억 및 80억 매개변수 모델 - 01:31 4. 오픈소스 AI의 중요성 - 01:54 5. 실제 활용 사례 - 03:45 6. 개발자 교육 계획 - 06:11 7. 메타의 생태계 구축 전략 - 07:19 8. 오픈소스 AI의 사회적 영향 - 09:53 9. 개발자 커뮤니티와의 소통 - 10:29 10. 보안 및 안전성 문제 - 12:00 11. 오픈소스 AI와 국가 안보 - 15:35 12. 오픈소스 AI의 혁신과 경제 성장 가속화 - 17:01 13. 애플과의 경쟁 및 폐쇄 시스템 문제 - 20:03 14. Llama 4의 미래 계획 - 24:01 15. AI 및 AGI의 장기적 비전 - 27:00 16. 제품 비전: 다양한 AI 에이전트 - 27:11 ㄴ 소규모 비즈니스 및 크리에이터를 위한 AI 에이전트 - 28:03 ㄴ AI 에이전트의 창작과 활용 - 30:11 17. AI에 대한 회의론 극복 - 31:29 18. 오픈소스 접근 방식의 중요성 - 32:00 19. AI와 경제 발전의 연관성 - 34:01 20. 웹 2.0과 AI의 발전 비교 - 34:12 21. 소셜 미디어와 AI의 미래 - 34:45 22. 인터뷰 종료 - 35:06
1. Opening - 00:00 2. Overview of Meta’s AI presentation - 00:39 3. Llama 3.1 model released - 00:53 ㄴ 405 billion parameter model - 01:00 7 and 8 billion parameter models - 01:31 4. Importance of open source AI - 01:54 5. Real-life use case - 03:45 6. Developer training plan - 06:11 7. Meta’s ecosystem construction strategy - 07:19 8. Social impact of open source AI - 09:53 9. Communication with the developer community - 10:29 10. Security and safety issues - 12:00 11. Open source AI and national security - 15:35 12. Open source AI accelerates innovation and economic growth - 17:01 13. Competition with Apple and closed system issues - 20:03 14. Llama 4's future plans - 24:01 15. Long-term vision of AI and AGI - 27:00 16. Product Vision: Diverse AI Agents - 27:11 ㄴ AI agents for small businesses and creators - 28:03 ㄴ Creation and use of AI agents - 30:11 17. Overcoming skepticism about AI - 31:29 18. Importance of an open source approach - 32:00 19. Connection between AI and economic development - 34:01 20. Comparison of the development of Web 2.0 and AI - 34:12 21. The future of social media and AI - 34:45 22. End of interview - 35:06
Mate this is incredible. 700 subs and you got THIS with ZUCK. I'm blown away, I'm really trying to start my own AI / Automation business and keep hitting walls of giving up my career as an HR Executive at a global company, the income, failure etc - you've shown me just what can be done. Legend.
In major label music promotion there are two roads promo managers know they can go by (...but in the long run there's still time to change the road you're on?). A. Give it away, give it away, give it away now...but then pull back on your fishing rod and ask people to "donate" or micropay on every "unit". This is when you've achieved a torrent of distribution from "giving it away". And B. License the snot out of it. Invoke IP lawyers. Pay dues in a policing agent (RIAA) that records every recordings issue (Archivist: Discogs). So what I perceive that Zuckerberg is doing is adopting the "give it away"...but if done right he can still make out in a big way.
Great questions around open and local LLMs. Open source AI will enable more innovation, simply because companies won't fear where they're sending thier data and maintain peace of mind when building new AI-native tech for customers.
Well put, Zuck! Excited for Llama 3.1 (all models) I like your acceptance that maybe we could have done social media a little bit better 🤣 (although hopefully Threads will help with that).
Rowan Cheung is the Founder of the world's largest AI newsletter, The Rundown AI. In just over a year, The Rundown has gained a following of over 600,000 subscribers and millions of monthly website visitors, grown primarily through organic word-of-mouth, who rely on the information provided by the newsletter daily.
idk what got into zuck but you gotta respect it ... if you want to earn back the trust of the public make as much as possible open source i hope meta inspires many other companies to be as transparent as possible
Zuck understands humanity of the 21st century, my take though. It's like making O2 available for everyone instead of refining it and bottling for profit.
Mark is a legend! There is no way open sourcing will make more money for Meta than keeping it private. He actually thinks it is better for the world and future of civilisation. Huge respect.
how T F this looks so realistic .... if i didnt read the comments i wouldnt have known that this video is AI generated... AI generated videos cant be this long... how you managed to have it so long?
Meta still needs to catch up with Google regarding developers, quality, ecosystem, etc. And also meta apps need to improve for users and what they really want, and getting rid of junk fake user accounts.
For those with short memories, until 2022 Zuck was saying the future of the world lay in the metaverse, throwing billions into a seemingly bottomless pit with little to show for it. Then, seemingly from nowhere, AI and ChatGPT came along & hit him smack bang in the face! OUCH!! And suddently metaverse, the technology he was so sure is the future he had changed the name of the entire company, was the last thing he wanted to talk about. People think these tech founds have got super-human vision. NO. They're just incredibly lucky people. Zuck literally has been saved by the AI boom. He was going entirely in the wrong direction, and his pride was not going to let him admit it. His obsession with metaverse was going to bankrupt Meta, with the irony of it all sticking like a sore thumb
Insightful discussion on AI with Mark Zuckerberg! SmythOS provides a secure and open platform to experiment with AI agents like Llama 3.1. #SmythOS #AIInnovation
Tried now, the 70B and 8B, very good. the 70B feels like GPT 3-5 or even GPT4oMini. The 405B actually has problems on Groq. By the way the 405B will be not possible to run on private GPUs...
Bright side! There are so many things to fix in this society: the relation to the animals, the acceptance of our own decline and end, the care for life in its diversity, ... Opening the technology sounds as a step in a better direction.
Ya but you need a supercomputer worth tens of thousands of dollars to run llama 3.1, but still some smaller organizations can have access to this is awesome
I like how mark stil ack like a regular person ..he dont cares wat people think ,even tho he is a legend and a life changer he and stuck up he kool on my book 💪
