Fantastic series, and the repeated mention of missionaries vs mercenaries in this episode is more relevant than ever.

7:00 Great to see how an expert user still has to go back and forth an debug it. Zapier is an engineering-centric company and so they are used to shipping things that need a lot of testing and back and forth. Important to have that mentality I've learned. Doesn't really translate for everyday average users in my opinion. Requires a nerdy technical marketers/ops person/developer to really harness it's power.

6:20 is a key point on how this helps avoid/fix broken Zap steps

Great post!

❤

Their Noise-Assassin technology is so superior, I think there is no other brands selling such technology today

🎉❤🎉

Amazing story and insights!

"Maybe you do, I can't do that" 2:39 😆😆

❤

How naive. Can't you see that Jensen Huang doesn't care about applications, as long as they require its GPUs to be built? Watch the interview at the Stanford Graduate School of Business: "almost everything we do, we create technology, we create markets"...

I just love this guy. He seems to be a wonderful person, so human, very smart and capable. Recently I have been using several of his github language model repositories. I bought a Linux x86 box and a used NVIDIA RTX 6000, really just to learn about this new field. Andrej has done so much to make this mind-bending technology understandable -- even for an old timer like me. Transformer systems are the first utterly new and commercially viable development in basic computer science since the 1960s. Obviously since then we have acquired amazingly fast CPUs capable of addressing huge amounts of RAM, as well as massive nonvolatile storage. But until these transformer models came along, the fundamental concept of data processing systems had not changed for decades. Although these LLMs are still being implemented within the Von Neumann architecture (augmented by vector arithmetic) they are fundamentally new and different beasts.

Americans hate to admit it but we've been under such a closed system for so long to the point our public education stopped enriching young minds 40 yrs ago and flipped into recruiting for agency & institutions taking away the most productive years from our workforce. It's been 90 years since we was in a very open private sector individual owners and operator creative posture. Everyone is a cog in the wheel more or less. We have so many grandfathered in economic middle men over the many phases of steam engine until today. Unlike most of the world that only individualized the past 50 years after the transitor age it leaves the west and America with many extra left over obstacles. It also leaves debts paid up front that have helped us get to this technology. Since I've retired and lived through all that 1900s, structuralism costs classical American decendants its only fair to remind everyone what these are from random Joe's perspective. Family birthrates ,18-30 year Olds trained up and entering workforce at the most creative and productive ages ( which by default tends to marry & help Maintain the elusive American prosperity) and the lack or loss of 31 -50 year old owner operators of local American infrastructure. Yes it's an unsustainable theme here that's been a very hefty price in building out our world over the 80 years of the transitor age. If any sectors are handed advantages in this new paradigm infrastructure, these are the ones who have paid the ultimate cost in my lifetime. New paradigm infrastructure where balance is there for better quality of life and at minimum restoration of all that's been compromised. We have so antiquated ways of doing things . Our city's are still under top down rule prohibition era reformed control mechanisms

We sure down want the woke AI road 👍

This agentic workflow sounds very much like parsing through iterative phases of collaborative problem solving, that is assumed to be captured in the massive data sets parsed by the networks.

Papers mentioned: Reflection "Self-Refine: Iterative Refinement with Self-Feedback, Madaan et al. (2023)" "Reflexion: Language Agents with Verbal Reinforcement Learning, Shinn et al. (2023)" Tool use "Gorilla: Large Language Model Connected with Massive APIs, Patil et al. (2023)" "MM-REACT: Prompting ChatGPT for Multimodal Reasoning and Action, Yang et al. (2023)" Planning "Chain-of-Thought Prompting Elicits Reasoning in Large Language Models, Wei et al. (2022)" "HuggingGPT: Solving AI Tasks with ChatGPT and its Friends in Hugging Face, Shen et al. (2023)" Multi-agent collaboration "Communicative Agents for Software Development, Qian et al. (2023)" "AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation, Wu et al. (2023)"

What's the difference between "Agent" and "Agentic workflow"? Will Agent include agentic workflow?

His example of successful AI implementation would involve exacly 0 AI. 🤔

He still not explain why he gave b to that poor kid

Thanks for sharing!

😂

Elon will struggle with scale

Love the analogy of the proverbial manager checking every 5 mins after assigning a task 🙃

Amazing talk :) Excited to see Mistral in action in Taskade's next Multi-Agent update! 🌈

I came here to learn how to invest after listening to a guy on radio talk about the importance of investing and how he made $989,000 in 2 months from $100k, somehow this video has helped shed light on some things, but I'm still confused, I'm a newbie and I'm open to ideas.

I'm watching you from age 16, GOAT AI teacher!!!!

Does Sequoia focus on candidate experience platform ?

Thank you

Now instead of farmers imagine what’s about to happen to millions of software engineers. These people love this so much because it will empower the business people to cut engineering completely out of the process of creating new products.

Sooo so amazing.. I could create a whole musical in a week or two at most with this 🎉

I constantly see figures who are big shots at funds like Sequoia who seem to have never done anything. They just graduated from prestigious universities then end up with a bunch of 'investor in' and 'board member of' listings on their LinkedIn experience history. Of all the entrepreneurs places like Sequoia, Y-Combi etc could hire, why do they promote people who have zero experience making something to pontificate to people about how to get things done. There's just something off about someone exiting university and going directly to guru and money allocator status.

Karpathy's Views on the Future of AGI Karpathy believes that AGI is now within sight and that various entities are working on building large language model (LLM) operating systems. These LLM operating systems will have customizable peripherals for different modalities like text, images, and audio, and will be connected to existing software infrastructure. Karpathy emphasizes the importance of building and influencing AGI to ensure its positive development. The Future of the AI Ecosystem OpenAI is seen as a dominant player in the AI ecosystem, but Karpathy suggests that there will be opportunities for other companies to build independent applications on top of the LLM operating system. The future of the AI ecosystem may resemble the early iPhone app ecosystem, with a variety of applications being developed over time. Karpathy draws a parallel between the current AI landscape and the operating system market, with a few proprietary systems and a long tail of open-source distributions. He highlights the importance of fully open-source LLMs, as opposed to just releasing binaries, for enabling true fine-tuning and customization. Factors Affecting the Training of Large Language Models Scale is the most important factor in training large language models, but other factors such as infrastructure, algorithms, and data quality also matter. There is a gap in the energy efficiency of running large language models compared to the human brain. There is a need to adapt computer architecture to the new data workflows and to improve precision, sparsity, and the von Neumann architecture. Elon Musk's Management Style and Views on AI Elon Musk's unique management style involves running small, strong, highly technical teams and being friendly to getting rid of low performers. He encourages a vibrant work environment where people are actively engaged and not afraid to leave unproductive meetings. Unlike traditional CEOs, Elon Musk is directly connected to the engineering team and spends a significant amount of time talking to them to understand the actual state of things and remove bottlenecks. He is deeply invested in the AI ecosystem and wants it to be healthy and vibrant, with many thriving startups. Elon Musk believes that traditional code is very composable, while neural networks are less composable by default. However, composability can be achieved through pre-training small pieces of the network and fine-tuning them as a whole. Limitations and Future Directions of AI Models Current AI models are very good at imitation learning but not so good at reinforcement learning or self-practice. AI models need to be able to practice and learn on their own in order to improve their performance. AI models need to be able to manipulate and reframe knowledge in order to truly understand it. Openness and Collaboration in the AI Ecosystem Open-source development will continue to keep pace with closed-source development as long as companies like Meta continue to release open-weights models and foster the ecosystem. The AI ecosystem would be cooler and more vibrant if there was more openness and collaboration, and if people were more willing to share what they have learned. Beyond the Transformer Neural Network Model The next big performance leap from models will require more than just modifying the Transformer architecture, and a new fundamental building block may be needed. While Transformer is an incredible model, it is unlikely to be the final neural network architecture. There is still potential for significant advancements in autoregressive and diffusion models, as well as in the co-design of hardware and network architectures. Encouraging a Vibrant AI Ecosystem Founders and builders in the AI field should focus on creating a vibrant ecosystem of startups and ensuring that startups can continue to compete with big tech companies.

Love the session. And Pat was being very euphemistic. By "replacing services with software", it really means "replacing human workers with software"

Thank you so much for sharing this!

I cannot understand how one can become so smart as Karpathy

The AI legend, I can't forget how easy was learning complext deep learning stuff by just taking Andrew Ng's courses

Hey fellow coders out there ! Any implementation of this you would recommend ? I used LangChain agents and it works pretty well.

Thank you so much for uploading this and not hoarding this knowledge in silicon valley. I think you are doing the world a massive favour!

Guys! Make sure to check out the papers he lists at 11:07! It is required reading for the exam 🙂

Andrej seems like such a good dude. Great moderation as well.

Could multiple LLMs be setup as Agents? Why hasn't anybody developed a system/app that takes the API's from the top LLMs, created agents for each, and then have these agents all work together to brainstorm, debate, review, and solve problems? I often get 4 different answers from 4 LLMs, so why not have them all setup as agents "in one room" working together to come up with the "best" solution. I can't find anybody that's tried this... why not? Wouldn't having the "top minds" (LLMs) working together produce better results?

I LOVE AI Agents... but I'm left wondering: Why hasn't anybody developed a system/app that takes the API's from the top LLMs, created agents for each, and then have these agents all work together to brainstorm, debate, review, and solve problems? I often get 4 different answers from 4 LLMs, so why not have them all setup as agents "in one room" working together to come up with the "best" solution. I can't find anybody that's tried this... why not? Wouldn't having the "top minds" (LLMs) working together produce better results?

Clip #2 [4:26-6:16] I love how Dr Ng humbly describes his work as laying down another brick on the golden road to AGI.

Certainly!

Take it easy on the coke big guy

Another fantastic video that blew me away!

The NPC dialogue still sounds robotic

Damn the next generation RPGs are gonna be wild.

Balancing autonomy and human intervention in AI agents is crucial. How will this shape future UX designs?

Amazing!