KUNGFU.​AI
KUNGFU.​AI
  • Видео 76
  • Просмотров 24 087
Hidden Layers: Decoded (OpenAI o3, Google Genie 2, DeepSeek-V3, Antibody Eng., MiniMax & more)
In this episode of Hidden Layers: Decoded, we dive into cutting-edge AI advancements over the last month. Explore Agentic AI and innovations like DeepMind Genie 2 and Cosmos Text2World, transforming virtual environments. Discover breakthroughs like RStar Math and DeepSeek v3, delivering efficiency and performance in reasoning and problem-solving. We also discuss test-time scaling, coding agents, and the drama behind the NeurIPS Best Paper Award.
Resources:
OpenAI o3 en.wikipedia.org/wiki/OpenAI_o3
DeepSeek-V3 www.deepseek.com/
Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction arxiv.org/abs/2404.02905
Abstract and Reasoning Corpus for Artificial General Intell...
Просмотров: 256

Видео

A Year of Hidden Layers: Blooper Edition
Просмотров 20День назад
A year ago, we decided to start a podcast. Where we ready when we started? Not even close. But here’s the thing: you can’t wait for perfect conditions to begin something worthwhile. You dive in, learn, and adapt. What better way to celebrate the one-year anniversary of “Hidden Layers” than with an embarrassing blooper reel from our last episode. Thanks to everyone who’s been along for the ride:...
Hidden Layers: AI Year in Review - Key Moments, Hot Takes, and 2025 Predictions | EP. 34
Просмотров 173Месяц назад
In this special 2024 AI Year in Review, Ron is joined by AI experts ZZ Si (Co-Founder & Distinguished Engineer), Emma Pirchalski (AI Strategist), and Michael Wharton (VP of Engineering) to reflect on the most important AI moments of 2024. They come together to discuss the defining stories, key breakthroughs, and major challenges that shaped AI in 2024. Ron leads the conversation, drawing out th...
Hidden Layers: Decoded (AlphaFold 3, AI Scaling, Robotic Surgery, LLaMA-Mesh, Gen AI Gaming & more)
Просмотров 187Месяц назад
In this episode of Hidden Layers: Decoded, Ron Green teams up with KUNGFU.AI's ZZ Si and Michael Wharton to explore groundbreaking advancements in artificial intelligence. From DeepMind’s AlphaFold 3 revolutionizing computational biology to debates on the limits of scaling AI models, the conversation covers all the latest AI news from the last month. Highlights include robotic surgery advanceme...
Is AI About to Replace Software Developers? Insights from Anaconda’s Peter Wang | EP. 32
Просмотров 2602 месяца назад
In this episode of "Hidden Layers," Ron Green dives into the transformative impact of AI on software development with Peter Wang, Chief AI and Innovation Officer at Anaconda. They discuss the rise of AI coding assistants, tools like GitHub Copilot and Cursor, and their potential to change how developers work. From coding support to the future of AI-native languages, they explore whether AI coul...
Hidden Layers: Decoded (Meta's Llama 3.2, OpenAI, Microsoft, Nobel Prize Reactions, GraphRAG & more)
Просмотров 2713 месяца назад
In this episode of Hidden Layers, Ron Green is joined by KUNGFU.AI's Michael Wharton and Dr. Steve Kramer to discuss the latest news in AI. They cover OpenAI’s leadership turnover, the rise of smaller, more efficient AI models, and the growing importance of AI governance. Plus, they explore Meta's Llama 3.2, a new multimodal model, and share insights from recent AI conferences. The conversation...
Cheating, Challenges, and Change: AI and Higher Education | EP. 30
Просмотров 5784 месяца назад
In this episode of Hidden Layers, host Ron Green talks with Cole Camplese, the Chief Information Officer at the University of Texas at Austin, about the transformative impact of AI in higher education. With over 25 years of experience in driving digital transformation, Cole discusses how AI is reshaping universities, from personalized learning to addressing the digital divide. They explore the ...
Revolutionizing Medicine: Custom Therapies for Your DNA
Просмотров 194 месяца назад
Discover how advancements in genomic-based therapy can transform drug development. By turning biology into an information science, we can create personalized treatments tailored to your DNA, improving health outcomes and saving lives. Learn about the future of medicine today! #GenomicTherapy #PersonalizedMedicine #DrugDevelopment #HealthInnovation #BiologyMeetsTech #FutureOfMedicine #CustomTher...
Hidden Layers: Decoded (OpenAI o1, DeepMind AlphaProteo, Apple Intelligence, Waymo & more) | EP.29
Просмотров 1784 месяца назад
In this episode of Hidden Layers, Ron Green and Michael Wharton dive into the latest advancements in artificial intelligence, with news from DeepMind, OpenAI, Waymo, Apple, and more. They discuss OpenAI's rumored Strawberry initiative, which promises enhanced reasoning capabilities for ChatGPT, the groundbreaking potential of Alpha Proteo for computational biology, and new developments in auton...
Client Testimonial from Dave Hogan, Chief Technology Officer, Guideline.ai
Просмотров 404 месяца назад
In this insightful testimonial, Dave Hogan, CTO of Guideline.ai, shares his experience working with KUNGFU.AI. He highlights how our collaboration helped streamline their processes, enhance decision-making with AI-driven solutions, and accelerate business growth. Hear firsthand how our tailored AI strategies supported Guideline.ai in overcoming challenges and achieving measurable results.
What Excites You Most About AI and Computer Vision? 🤖✨
Просмотров 574 месяца назад
In this video, we explore the thrilling advancements in AI and computer vision. From multi-modal models to robotics, discover the cutting-edge techniques shaping our future and how they can transform various industries. Join us in this enlightening discussion! #AI #ComputerVision #TechInnovation #Robotics #MachineLearning #FutureTechnology #DataScience #VisionaryTech #MultiModalAI #IntelligentS...
Beyond Generative AI: Rediscovering Computer Vision’s Legacy | EP.28
Просмотров 974 месяца назад
Beyond Generative AI: Rediscovering Computer Vision’s Legacy | EP.28
KUNGFU.AI Presents: Are Emergent Abilities of LLMs a Mirage?
Просмотров 944 месяца назад
KUNGFU.AI Presents: Are Emergent Abilities of LLMs a Mirage?
The Future of AI: Training Models, High-Quality Data, and Substantial Computing Power
Просмотров 265 месяцев назад
The Future of AI: Training Models, High-Quality Data, and Substantial Computing Power
Hidden Layers: Decoded (Meta's Llama 3.1, DeepMind's AlphaProof, SAM2, FLUX.1, BRAG, & more) | EP.27
Просмотров 2315 месяцев назад
Hidden Layers: Decoded (Meta's Llama 3.1, DeepMind's AlphaProof, SAM2, FLUX.1, BRAG, & more) | EP.27
AI and the Evolution of Open Source: Insights from NVIDIA's Dr. Katrina Riehl | EP.26
Просмотров 1356 месяцев назад
AI and the Evolution of Open Source: Insights from NVIDIA's Dr. Katrina Riehl | EP.26
Hidden Layers: Decoded (Microsoft’s Florence-2, GraphRAG, Claude 3.5, NVIDIA’s NIM and more) | EP.25
Просмотров 2176 месяцев назад
Hidden Layers: Decoded (Microsoft’s Florence-2, GraphRAG, Claude 3.5, NVIDIA’s NIM and more) | EP.25
3 Lessons From 6 Years in the AI Trenches | EP. 24
Просмотров 1127 месяцев назад
3 Lessons From 6 Years in the AI Trenches | EP. 24
Hidden Layers: Decoded (Mamba-2, KANs, OpenAI's Superalignment Team, and more) | EP.23
Просмотров 3627 месяцев назад
Hidden Layers: Decoded (Mamba-2, KANs, OpenAI's Superalignment Team, and more) | EP.23
AI's Secret Foundation: High Performance Computing | EP.22
Просмотров 1197 месяцев назад
AI's Secret Foundation: High Performance Computing | EP.22
Scott Aaronson: Aligning Superintelligent AGI | EP.21
Просмотров 2 тыс.8 месяцев назад
Scott Aaronson: Aligning Superintelligent AGI | EP.21
Open Source vs Closed Models: Who Leads the LLM Landscape?
Просмотров 408 месяцев назад
Open Source vs Closed Models: Who Leads the LLM Landscape?
Vision Language Models: Revolutionizing Image Analysis and Medical Diagnosis
Просмотров 1338 месяцев назад
Vision Language Models: Revolutionizing Image Analysis and Medical Diagnosis
Trendspotting in AI: Governance Strategies for Tomorrow | EP.20
Просмотров 878 месяцев назад
Trendspotting in AI: Governance Strategies for Tomorrow | EP.20
Running State-of-the-Art AI Models on a MacBook Pro: Impressive Performance and Specs
Просмотров 4798 месяцев назад
Running State-of-the-Art AI Models on a MacBook Pro: Impressive Performance and Specs
Unlocking the Magic: Microsoft Copilot Workspace and Bug Fixes
Просмотров 498 месяцев назад
Unlocking the Magic: Microsoft Copilot Workspace and Bug Fixes
Apple's MM1 Release: Shaping the Future of Algorithm Development
Просмотров 1148 месяцев назад
Apple's MM1 Release: Shaping the Future of Algorithm Development
Hidden Layers: Decoded (Phi-3, Llama 3, Transformer-XL, InternVL, Apple's MM1 and more) | EP.19
Просмотров 2598 месяцев назад
Hidden Layers: Decoded (Phi-3, Llama 3, Transformer-XL, InternVL, Apple's MM1 and more) | EP.19
Inside the AI Talent Hunt: Strategies for Success and Inclusion | Episode 18
Просмотров 1749 месяцев назад
Inside the AI Talent Hunt: Strategies for Success and Inclusion | Episode 18
Understanding Real-World Applications of Graph Neural Networks | Episode 17
Просмотров 1019 месяцев назад
Understanding Real-World Applications of Graph Neural Networks | Episode 17

Комментарии

  • @burnytech
    @burnytech 29 дней назад

    <3

  • @Dr.CandanEsin
    @Dr.CandanEsin Месяц назад

    While my external state tries to listen to you and follow the presentation, internal states wish to extrapolate the ideas and try to put them into understandable equilibrium. Thus, the Markov boundary cannot function properly. Especially when the presenter is reading a text instead of giving impromptu definitions and examples, Markov takes away his blanket from the stage totally 😏. Like when magician removes the blanket subject disappears. PS: Sigmund Freud proposed such a mathematical/metapsychological system, in his work known as Project a Scientific Psychology but not published until James Strachey included it posthumously. Jaak Panksepp also contributed largely with animal studies.

  • @9worlds20
    @9worlds20 Месяц назад

    I definitely disagree with the position that the human mind and its intellect are "in line" with our current understanding of physics. The body, hormones and emotions are inline with evolution, but intellect is the only factor that is, or can be alien in nature. AI will be an artificial way that our intellect expands; but does, or will it ever be able to have a psychedelic experience, or understand the hidden realities, that our consciousness touches. I guess the goal is that super intelligent, self creating and evolving AI will discover new physics for us and perhaps tap into hidden realities, if that is even possible. Maybe the super intelligent machines cannot use matter in a way to construct the machine needed to create exotic physics or such; thus it will expand into the mind and consciousness, which does not require physical matter in the same sense. Most likely we are facing: Super intelligent AI < to > human brain merging, with the aid of technology. Were AIs can learn and use the human brain.. like symbiots. The problem still remains of how to create the machines that will produce the exotic physics. Abstraction can't manifest matter out of thin air. Azvanir

  • @ericgreenberg9495
    @ericgreenberg9495 Месяц назад

    Reason for internal development is that the applications with roi do not exist and s/w development must be business process and results oriented. The packaged applications do not exist for solutions versus model use. The applications that go to market will emerge from custom development by consultancies and enterprises in partnership with certain vendors to begin a new landscape of AI solutions. I also built multimodal UX solutions as far back as 2015/16. The “search/model result” of today is a 1.0 low utility output without strong extensibility.

  •  3 месяца назад

    Thank you for the super-clear and layman-level explanation of a rather complex theory.

    • @kungfuai
      @kungfuai 3 месяца назад

      Happy you enjoyed it!

  • @YouTubeSEO-u2v
    @YouTubeSEO-u2v 4 месяца назад

    Awesome

  • @annieorben
    @annieorben 4 месяца назад

    Maybe OpenAI will invest in more efficient inference time computation? There are a few companies already focusing on this problem. I like the way Grok approaches it with massively parallel inference stored on CPU memory using sram and interconnects between all of the CPUs. It scales really well.

    • @kungfuai
      @kungfuai 4 месяца назад

      Great point! Optimizing inference efficiency is increasingly critical as models continue to scale. Techniques like parallel computation, specialized memory architectures, and advanced interconnects are all being explored to reduce latency and increase throughput. Interestingly, increasing inference compute can enhance model performance, often more effectively than scaling model size or expensive retraining. OpenAI’s new projects, like Strawberry (Q*), explore advanced inference techniques such as self-talk, optimizing reasoning abilities through methods like best-of-N sampling and internal deliberation, reflecting a shift towards leveraging more compute at inference time.

  • @ryanhenning
    @ryanhenning 5 месяцев назад

    So good! I love questions like these.

    • @kungfuai
      @kungfuai 5 месяцев назад

      Thanks for the kind words, Ryan!

  • @silberlinie
    @silberlinie 5 месяцев назад

    eine Gemeinheiz, Scott auf einen so kleinen Stuhl sitzen zu lassen ...

  • @rigidrobot
    @rigidrobot 5 месяцев назад

    Not just dogs have aligned humans. wheat has aligned humans too.

  • @ramitube21
    @ramitube21 6 месяцев назад

    Excellent video.

    • @kungfuai
      @kungfuai 6 месяцев назад

      Thank you very much!

  • @mikaelfiil3733
    @mikaelfiil3733 7 месяцев назад

    A small comment on the detection of AI generated text and comments etc. Be careful to distinguish between an argument (any argument from any person) and the false attribution of that argument to a specific person. We don't want a thought police.

  • @JustLennyBenny
    @JustLennyBenny 7 месяцев назад

    Most people still think AI is not even as far as we are right now. Some even think its hardcoded responses >.> and they are hellbound on not believe anything other then that. Super ignorant people, but I like to proof them wrong soon. :)

  • @ArtII2Long
    @ArtII2Long 7 месяцев назад

    Think of A.I. as a psychopath clinically. A human psychopath can be guided and leveraged based on self interest. A.I. doesn't have self interest, it has instead it's directive and wil be dead set to complete it. So, every directive should include "Without causing us harm". With ASI, that directive might be understood.

  • @nikhilsulghur7589
    @nikhilsulghur7589 8 месяцев назад

    this is an succinct overview of Bayesian mechanics, thanks for the upload.

  • @jasonclarke2346
    @jasonclarke2346 8 месяцев назад

    very clear - thanks!

  • @UncleDavid
    @UncleDavid 10 месяцев назад

    lol u sound like Siri reading some gpt script

  • @dineshcuts
    @dineshcuts Год назад

    Source code ?

  • @southpointe63rd74
    @southpointe63rd74 Год назад

    Looking forward to hearing from you!

  • @DeskLampManifesto
    @DeskLampManifesto Год назад

    How do I sign up?

  • @onepunchman8721
    @onepunchman8721 Год назад

    Great presentation 😊 Can you tell what the differences are to CUTIE?