Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Day 15 of learning AI bs from scratch > looked into seq2seq models > understood the encoder-decoder architecture > implemented a basic Seq2Seq model using RNNs > how the decoder generates outputs one step at a time > read couple more chapters of > understood the fixed-context bottleneck I... show more

Arham Amin

2,414 subscribers

31,680 views • 1 month ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

WOW! I am floored and sort of scared. The Human Synapse Decoder (2 channel) via ZUNA AI translated brain waves to text while I was dreaming. This is the AI dream decoder text: “ancient library... whispers of forgotten knowledge” In the morning I woke up and recorded what I can remember from my dreams and here is that recoding played back. I had not seen the results of The Human Synapse Decoder yet. I really can’t understand how two channels could do this or believe this happened. This is day one of research and the AI researchers at The Zero-Human Labs directed by Mr. Grok is digging deep into papers and the secrets of the NeuroSky chips and proprietary systems. But I think we are on the way to the first AI Liminal State/Dream State Decoding technology and you folks are seeing it raw in real-time! Thank you!

WOW! I am floored and sort of scared. The Human Synapse Decoder (2 channel) via ZUNA AI translated brain waves to text while I was dreaming. This is the AI dream decoder text: “ancient library... whispers of forgotten knowledge” In the morning I woke up and recorded what I can remember from my dreams and here is that recoding played back. I had not seen the results of The Human Synapse Decoder yet. I really can’t understand how two channels could do this or believe this happened. This is day one of research and the AI researchers at The Zero-Human Labs directed by Mr. Grok is digging deep into papers and the secrets of the NeuroSky chips and proprietary systems. But I think we are on the way to the first AI Liminal State/Dream State Decoding technology and you folks are seeing it raw in real-time! Thank you!

Brian Roemmele

396,208 views • 7 months ago

the thing you rent for $200 a month just became something you can own for $1,700 once but the money is not even the real story for the first time a 200 billion parameter model is not in a datacenter, it is sitting on a desk the cloud spent years convincing you a model this size needed their servers, their meter, their monthly bill people are stacking four subscriptions into a $440 a month bill to rent what one box this size now owns outright it needed a box the size of a book the moment the model moves from their datacenter to your desk, the whole game changes it stops being about who has the best AI it becomes about who ships it on every desk the cloud told you this needed a datacenter it needed a desk i did the full math on what this kills in the article below

the thing you rent for $200 a month just became something you can own for $1,700 once but the money is not even the real story for the first time a 200 billion parameter model is not in a datacenter, it is sitting on a desk the cloud spent years convincing you a model this size needed their servers, their meter, their monthly bill people are stacking four subscriptions into a $440 a month bill to rent what one box this size now owns outright it needed a box the size of a book the moment the model moves from their datacenter to your desk, the whole game changes it stops being about who has the best AI it becomes about who ships it on every desk the cloud told you this needed a datacenter it needed a desk i did the full math on what this kills in the article below

John Doe

26,096 views • 3 months ago

We released physics-intern: a simple harness for science problems! It gets models like Gemini 3.1 Pro to go from 17.7 -> 31.4, thus beating GPT 5.5 Pro. The physics-intern harness can wrap any model and via dedicated subagent boost the performance of the vanilla reasoning models. While I think more and more of these harness capability gains will be absorbed into the models (like prompting tricks disappeared over time) there is a lot to be gained right now by building good scaffolds for those models and integrating tools well. Interestingly, the exception we found that GPT 5.5 Pro actually didn't benefit from the physics-intern harness! Read more about it here: PS: I think the Harness[Model] notation is kind of nice.

We released physics-intern: a simple harness for science problems! It gets models like Gemini 3.1 Pro to go from 17.7 -> 31.4, thus beating GPT 5.5 Pro. The physics-intern harness can wrap any model and via dedicated subagent boost the performance of the vanilla reasoning models. While I think more and more of these harness capability gains will be absorbed into the models (like prompting tricks disappeared over time) there is a lot to be gained right now by building good scaffolds for those models and integrating tools well. Interestingly, the exception we found that GPT 5.5 Pro actually didn't benefit from the physics-intern harness! Read more about it here: PS: I think the Harness[Model] notation is kind of nice.

Leandro von Werra

97,519 views • 4 months ago

“How I Get” is out today!! & you can now pre-order / pre-save my deluxe album, A Matter of Time: The Final Hour, out April 10th…at the link below xx The Final Hour picks up where A Matter of Time left off… the light is softer, the air like spring after the storm, stepping into a new emotional chapter 🪜🤍 I am so excited for you guys to hear it!!

“How I Get” is out today!! & you can now pre-order / pre-save my deluxe album, A Matter of Time: The Final Hour, out April 10th…at the link below xx The Final Hour picks up where A Matter of Time left off… the light is softer, the air like spring after the storm, stepping into a new emotional chapter 🪜🤍 I am so excited for you guys to hear it!!

laufey

168,294 views • 7 months ago

I made a step by step one-hour tutorial on how to make a VR shooter in Unity! After releasing the first 3-parts of my series on how to make a VR game in Unity 6.2, I wanted to build a VR game from scratch using everything we’ve learned so far. Out of 10 different themes, the one that won the Patreon poll was… a VR Shooter! Here’s the result of that challenge: • A VR rig with climbing locomotion • A pistol with silencer • Guard patrols • A guard AI vision system that can detect the player If you’d like to watch the full tutorial and get the Unity project files, it’s available right now on Patreon : 👉

I made a step by step one-hour tutorial on how to make a VR shooter in Unity! After releasing the first 3-parts of my series on how to make a VR game in Unity 6.2, I wanted to build a VR game from scratch using everything we’ve learned so far. Out of 10 different themes, the one that won the Patreon poll was… a VR Shooter! Here’s the result of that challenge: • A VR rig with climbing locomotion • A pistol with silencer • Guard patrols • A guard AI vision system that can detect the player If you’d like to watch the full tutorial and get the Unity project files, it’s available right now on Patreon : 👉

Quentin Valembois

11,449 views • 11 months ago

i just ran Google's brand new Unsloth Gemma4 12B dense GGUF on my RTX 4060 using llama.cpp + CUDA 13.2 21 tokens per second. on a budget consumer GPU. locally. no API. no cloud. no subscription. and the benchmarks are absolutely cooked # first let's talk architecture because this is genuinely different every multimodal model you've used has a frozen vision encoder + frozen audio encoder + LLM backbone glued together Gemma 4 12B is different it's a single decoder only transformer. that's it. vision? raw 48×48 pixel patches → one matmul → projected directly into the LLM audio? raw 16kHz signal sliced into 40ms frames → linear projection → same LLM input space no encoder tax. no latency penalty. no fragmented memory to put the encoder savings in perspective: old Gemma 4 26B approach: - 550M param vision encoder (frozen) - 300M param audio encoder (frozen) - LLM backbone Gemma 4 12B: - 35M param vision embedder (a single matmul) - no audio encoder at all - LLM backbone handles EVERYTHING 550M → 35M for vision alone. that's a 15x reduction this is why the gemma-4-12b-it-Q4_K_M.gguf is just 6.6 GBs!!! and it has 256K native context context # Benchmarks: AIME 2026 (math olympiad): 77.5% GPQA Diamond (expert science): 78.8% LiveCodeBench v6 (real code): 72% Codeforces ELO: 1659 MMLU Pro: 77.2% MATH-Vision: 79.7% BigBench Extra Hard: 53% inference → llama.cpp, LM Studio, vLLM, SGLang llamacpp flags: -m "gemma-4-12b-it-Q4_K_M.gguf" -ngl 99 -c 8000 -v --port 8080 Available on huggingface now! Link below

Alok

281,007 views • 3 months ago

The recent Massachusetts Institute of Technology (MIT) CSAIL paper published on Recursive Language Models is a fascinating look into how AI systems reason in 2026. You can check it out here ➡️ → The paper notes that even frontier models suffer from “context rot” as inputs grow longer. More tokens don’t mean more understanding. Instead of compressing or summarising, RLMs “treat long prompts as part of an external environment” …and let the model programmatically inspect, decompose, and recursively requery itself over precise sections. This matters in Web3. Why? Smart contracts are long, stateful, and brittle. One missed assumption = unnecessary hassle. In the video below, we applied RLM principles to improve the prompt shown. The old prompt would try to force multi-step pauses. Therefore breaking the AI's workflow. Now it decomposes requirements, scores confidence for each component, verifies the logic, and then synthesises. Result = Production-ready contracts with flagged risks, not blind single-pass outputs. Try it out with our Smart Contract Generator today!

The recent Massachusetts Institute of Technology (MIT) CSAIL paper published on Recursive Language Models is a fascinating look into how AI systems reason in 2026. You can check it out here ➡️ → The paper notes that even frontier models suffer from “context rot” as inputs grow longer. More tokens don’t mean more understanding. Instead of compressing or summarising, RLMs “treat long prompts as part of an external environment” …and let the model programmatically inspect, decompose, and recursively requery itself over precise sections. This matters in Web3. Why? Smart contracts are long, stateful, and brittle. One missed assumption = unnecessary hassle. In the video below, we applied RLM principles to improve the prompt shown. The old prompt would try to force multi-step pauses. Therefore breaking the AI's workflow. Now it decomposes requirements, scores confidence for each component, verifies the logic, and then synthesises. Result = Production-ready contracts with flagged risks, not blind single-pass outputs. Try it out with our Smart Contract Generator today!

ChainGPT

82,385 views • 8 months ago

A Closer Look at Alchemist AI v3.5🔎 In v1, one agent handled everything. You’d enter a prompt like “make a snake game” and get a single block of code—one input, one output. v2 introduced multiple agents with specialized roles: prompt refinement, frontend, backend, and review. It brought more structure, but the process was still linear. Each step produced a single output. With v3, a single agent could handle multiple tasks. It generated HTML, CSS, JavaScript, searched for assets, and more—all in one go, while maintaining full context. Now in v3.5, we combine both models. Multiple agents work in parallel, each capable of producing multiple outputs at once. The prompt compiler adapts based on what you're building. A 3D game? It prioritizes game engines, rendering, and mechanics. A website or 2D app? It shifts focus to relevant frameworks and tools. The tool handler can call several services simultaneously, feeding context-aware data into the code generator for real-time execution. From single-output generation to parallel, adaptive workflows. In the next breakdown, we’ll dive into the architecture behind our proprietary engine in v3.5.

A Closer Look at Alchemist AI v3.5🔎 In v1, one agent handled everything. You’d enter a prompt like “make a snake game” and get a single block of code—one input, one output. v2 introduced multiple agents with specialized roles: prompt refinement, frontend, backend, and review. It brought more structure, but the process was still linear. Each step produced a single output. With v3, a single agent could handle multiple tasks. It generated HTML, CSS, JavaScript, searched for assets, and more—all in one go, while maintaining full context. Now in v3.5, we combine both models. Multiple agents work in parallel, each capable of producing multiple outputs at once. The prompt compiler adapts based on what you're building. A 3D game? It prioritizes game engines, rendering, and mechanics. A website or 2D app? It shifts focus to relevant frameworks and tools. The tool handler can call several services simultaneously, feeding context-aware data into the code generator for real-time execution. From single-output generation to parallel, adaptive workflows. In the next breakdown, we’ll dive into the architecture behind our proprietary engine in v3.5.

ALCHEMIST AI 🔮

12,689 views • 1 year ago

While building SEV0, I ran into moments where natural language prompting just wasn't enough - too vague, too indirect. So I started using a method that bridges that gap. I call it Bridged Prompting - a technique where you temporarily step out of the prompt-response loop to manipulate something directly using an AI-generated UI, then step back in. Think of it like GenUI, but more user-driven and transient. AI generates a UI on-the-fly, tailored to your prompt, and lets you manipulate the artifact directly before resuming the conversation. Sure, you could build a full tool, switch tabs, wire it up and round-trip your data. But with Bridged Prompting, the AI just spins up a temporary interface right in the flow. In the video below, I used it to construct the hallway system to closely match the layout of the severed floor. This is something I needed to do for this one project, this one time - I didn't need to make a whole separate app. Bridged Prompting lets you: - Make precise edits visually or structurally - "Hit save" to persist changes to local storage or a backend - Return to your natural language prompt

While building SEV0, I ran into moments where natural language prompting just wasn't enough - too vague, too indirect. So I started using a method that bridges that gap. I call it Bridged Prompting - a technique where you temporarily step out of the prompt-response loop to manipulate something directly using an AI-generated UI, then step back in. Think of it like GenUI, but more user-driven and transient. AI generates a UI on-the-fly, tailored to your prompt, and lets you manipulate the artifact directly before resuming the conversation. Sure, you could build a full tool, switch tabs, wire it up and round-trip your data. But with Bridged Prompting, the AI just spins up a temporary interface right in the flow. In the video below, I used it to construct the hallway system to closely match the layout of the severed floor. This is something I needed to do for this one project, this one time - I didn't need to make a whole separate app. Bridged Prompting lets you: - Make precise edits visually or structurally - "Hit save" to persist changes to local storage or a backend - Return to your natural language prompt

Chris Tate

42,558 views • 1 year ago

I don’t think people realize how big a jump Grok 4.6 is. Look at this comparison with Fable on the same prompt. They’re pretty close, but Grok took half the time at ONE TENTH the cost. And I prefer a lot of choices Grok made: - fills the hero image to the top instead of leaving an ugly white strip - flat sharp tiles instead of the usual rounded card slop - fewer pointless labels and elements, more restraint - lighter text, less dense and overwhelming Fable certainly has more depth if you’re trying to do crazy edge-of-distribution stuff. It’s a much bigger model. But for the majority of work, you can get Fable quality for a 90% discount. Just insane how much of a step up it is from Grok 4.5.

I don’t think people realize how big a jump Grok 4.6 is. Look at this comparison with Fable on the same prompt. They’re pretty close, but Grok took half the time at ONE TENTH the cost. And I prefer a lot of choices Grok made: - fills the hero image to the top instead of leaving an ugly white strip - flat sharp tiles instead of the usual rounded card slop - fewer pointless labels and elements, more restraint - lighter text, less dense and overwhelming Fable certainly has more depth if you’re trying to do crazy edge-of-distribution stuff. It’s a much bigger model. But for the majority of work, you can get Fable quality for a 90% discount. Just insane how much of a step up it is from Grok 4.5.

Anshu

54,593 views • 1 month ago

In my class, I teach the autoencoder by asking everyone to stand up. 🙆 Stretch your arms out wide. Imagine you are holding a heavy textbook (like Introduction to Algorithms by Prof. Cormen), the whole thing, every page. Now bring your hands slowly together until they almost touch your neck. The bottle "neck." The final exam is tomorrow and you are allowed one cheat sheet: whatever you can scribble on your palm. All nine hundred pages have to survive the squeeze. That is the encoder. Now imagine you sit in the exam. Push your arms back out to where they started. You try to rebuild the textbook from your palm notes. That is the decoder. Of course you cannot get every page back. What you get back is what mattered enough to write down, and the gap between the two is the loss the network is trying to shrink. I call it AI by Arms 🙆. It gets a laugh, and then it gets remembered. Goal: squeeze four numbers down to two, then rebuild the original four from them. 1. Given Let us start with four training examples: X1, X2, X3, X4. 2. Auto (copy to targets) We copy the training examples straight into the targets. That is the whole trick behind the name: "auto" is Greek for "self", and the data is its own label. 3. Encoder, layer 1 Let us multiply the inputs by the weights, add the biases, and apply ReLU. Negative values get crossed out and become zero. 4. Encoder, layer 2 (the bottleneck) We do it again, and now the four dimensions have become two. This layer is called the bottleneck, because everything has to fit through it. 5. Decoder, layer 1 Let us go back the other way: multiply, add, ReLU. This time there are no negatives to cross out. 6. Decoder, layer 2 We multiply once more and get the outputs Y. This is the decoder's attempt to rebuild the four original numbers from the two it was given. 7. MSE loss gradients Let us compare Y with the targets Y'. The gradient is 2 x (Y - Y'): subtract, then double. Those gradients kick off backpropagation, and the weights start to learn. Your entire education is all about encoding and decoding!

In my class, I teach the autoencoder by asking everyone to stand up. 🙆 Stretch your arms out wide. Imagine you are holding a heavy textbook (like Introduction to Algorithms by Prof. Cormen), the whole thing, every page. Now bring your hands slowly together until they almost touch your neck. The bottle "neck." The final exam is tomorrow and you are allowed one cheat sheet: whatever you can scribble on your palm. All nine hundred pages have to survive the squeeze. That is the encoder. Now imagine you sit in the exam. Push your arms back out to where they started. You try to rebuild the textbook from your palm notes. That is the decoder. Of course you cannot get every page back. What you get back is what mattered enough to write down, and the gap between the two is the loss the network is trying to shrink. I call it AI by Arms 🙆. It gets a laugh, and then it gets remembered. Goal: squeeze four numbers down to two, then rebuild the original four from them. 1. Given Let us start with four training examples: X1, X2, X3, X4. 2. Auto (copy to targets) We copy the training examples straight into the targets. That is the whole trick behind the name: "auto" is Greek for "self", and the data is its own label. 3. Encoder, layer 1 Let us multiply the inputs by the weights, add the biases, and apply ReLU. Negative values get crossed out and become zero. 4. Encoder, layer 2 (the bottleneck) We do it again, and now the four dimensions have become two. This layer is called the bottleneck, because everything has to fit through it. 5. Decoder, layer 1 Let us go back the other way: multiply, add, ReLU. This time there are no negatives to cross out. 6. Decoder, layer 2 We multiply once more and get the outputs Y. This is the decoder's attempt to rebuild the four original numbers from the two it was given. 7. MSE loss gradients Let us compare Y with the targets Y'. The gradient is 2 x (Y - Y'): subtract, then double. Those gradients kick off backpropagation, and the weights start to learn. Your entire education is all about encoding and decoding!

Tom Yeh

10,442 views • 14 days ago

In my class, I teach the autoencoder by asking everyone to stand up. 🙆 Stretch your arms out wide. Imagine you are holding a heavy textbook (like Introduction to Algorithms by Prof. Cormen), the whole thing, every page. Now bring your hands slowly together until they almost touch your neck. The bottle "neck." The final exam is tomorrow and you are allowed one cheat sheet: whatever you can scribble on your palm. All nine hundred pages have to survive the squeeze. That is the encoder. Now imagine you sit in the exam. Push your arms back out to where they started. You try to rebuild the textbook from your palm notes. That is the decoder. Of course you cannot get every page back. What you get back is what mattered enough to write down, and the gap between the two is the loss the network is trying to shrink. I call it AI by Arms 🙆. It gets a laugh, and then it gets remembered. Goal: squeeze four numbers down to two, then rebuild the original four from them. 1. Given Let us start with four training examples: X1, X2, X3, X4. 2. Auto (copy to targets) We copy the training examples straight into the targets. That is the whole trick behind the name: "auto" is Greek for "self", and the data is its own label. 3. Encoder, layer 1 Let us multiply the inputs by the weights, add the biases, and apply ReLU. Negative values get crossed out and become zero. 4. Encoder, layer 2 (the bottleneck) We do it again, and now the four dimensions have become two. This layer is called the bottleneck, because everything has to fit through it. 5. Decoder, layer 1 Let us go back the other way: multiply, add, ReLU. This time there are no negatives to cross out. 6. Decoder, layer 2 We multiply once more and get the outputs Y. This is the decoder's attempt to rebuild the four original numbers from the two it was given. 7. MSE loss gradients Let us compare Y with the targets Y'. The gradient is 2 x (Y - Y'): subtract, then double. Those gradients kick off backpropagation, and the weights start to learn. Your entire education is all about encoding and decoding!

In my class, I teach the autoencoder by asking everyone to stand up. 🙆 Stretch your arms out wide. Imagine you are holding a heavy textbook (like Introduction to Algorithms by Prof. Cormen), the whole thing, every page. Now bring your hands slowly together until they almost touch your neck. The bottle "neck." The final exam is tomorrow and you are allowed one cheat sheet: whatever you can scribble on your palm. All nine hundred pages have to survive the squeeze. That is the encoder. Now imagine you sit in the exam. Push your arms back out to where they started. You try to rebuild the textbook from your palm notes. That is the decoder. Of course you cannot get every page back. What you get back is what mattered enough to write down, and the gap between the two is the loss the network is trying to shrink. I call it AI by Arms 🙆. It gets a laugh, and then it gets remembered. Goal: squeeze four numbers down to two, then rebuild the original four from them. 1. Given Let us start with four training examples: X1, X2, X3, X4. 2. Auto (copy to targets) We copy the training examples straight into the targets. That is the whole trick behind the name: "auto" is Greek for "self", and the data is its own label. 3. Encoder, layer 1 Let us multiply the inputs by the weights, add the biases, and apply ReLU. Negative values get crossed out and become zero. 4. Encoder, layer 2 (the bottleneck) We do it again, and now the four dimensions have become two. This layer is called the bottleneck, because everything has to fit through it. 5. Decoder, layer 1 Let us go back the other way: multiply, add, ReLU. This time there are no negatives to cross out. 6. Decoder, layer 2 We multiply once more and get the outputs Y. This is the decoder's attempt to rebuild the four original numbers from the two it was given. 7. MSE loss gradients Let us compare Y with the targets Y'. The gradient is 2 x (Y - Y'): subtract, then double. Those gradients kick off backpropagation, and the weights start to learn. Your entire education is all about encoding and decoding!

Tom Yeh

22,703 views • 1 month ago

Scottish fans drank Boston dry? 🏴󠁧󠁢󠁳󠁣󠁴󠁿 🍺 We asked UpRock Semantic Video Search. Seconds later it nailed the exact timestamp. Not the title. Not a guess. The precise moment it happened. This is the new gap in AI. Most tools read transcripts or the metadata and quit. UpRock actually "watches" the video: what is said, what is shown, the action, the context, the vibe, why it matters. Last livestream we connected UpRock Video MCP to one of OpenAI's cheapest models. Without UpRock the model was blind. With it? The model spotted background signs, clothing details, objects, and live action like a pro. More and more of the internet's richest knowledge lives inside video. We are unlocking it. Over 100,000 videos indexed and climbing fast. Web agents learned to read. Now they are learning to watch. Watch the demo (Scottish fans moment): UpRock enables your AI to understand video the way humans do, saving you from slow, manual digging. Who else wants their AI to actually watch and save time?

Scottish fans drank Boston dry? 🏴󠁧󠁢󠁳󠁣󠁴󠁿 🍺 We asked UpRock Semantic Video Search. Seconds later it nailed the exact timestamp. Not the title. Not a guess. The precise moment it happened. This is the new gap in AI. Most tools read transcripts or the metadata and quit. UpRock actually "watches" the video: what is said, what is shown, the action, the context, the vibe, why it matters. Last livestream we connected UpRock Video MCP to one of OpenAI's cheapest models. Without UpRock the model was blind. With it? The model spotted background signs, clothing details, objects, and live action like a pro. More and more of the internet's richest knowledge lives inside video. We are unlocking it. Over 100,000 videos indexed and climbing fast. Web agents learned to read. Now they are learning to watch. Watch the demo (Scottish fans moment): UpRock enables your AI to understand video the way humans do, saving you from slow, manual digging. Who else wants their AI to actually watch and save time?

UpRock

75,157 views • 3 months ago

AI Is Moving Beyond “Generating Videos” — Toward “Generating Worlds” Over the past two years, AI video models have advanced at an astonishing pace. From Runway and Pika to Sora and Veo, AI-generated videos have become increasingly realistic and more consistent with the physical laws of the real world. Many people believe the next objective is simply to generate videos that are longer, sharper, and more lifelike. But if we take a step back, we can see that the real transformation is not happening in video itself. It is happening in world models. What Is a World Model? In 1943, psychologist Kenneth Craik proposed an idea that would influence artificial intelligence research for decades. He argued that the human brain does not merely react to the outside world. Instead, it maintains an internal model of how the world works. Because we have this internal model, we can predict the outcome of an action before we actually take it. Before crossing a road, we estimate whether a car will pass by. Before catching a ball, we predict its trajectory. These abilities come from continuously simulating the world in our minds, rather than relying entirely on trial and error. This idea later became known by a more formal term: World Model. A world model does not describe a single image or a fixed video clip. It is an internal representation capable of continuously simulating the rules and dynamics of the real world. Why Is AI Research Turning Toward World Models? Because predicting “what comes next” is becoming increasingly central to how AI systems work. Language models predict the next token. Image models predict the next step in the denoising process. Video models predict the next frame. A world model, however, attempts to predict something broader: What should the world look like in the next moment? In 2018, David Ha and Jürgen Schmidhuber proposed in their paper World Models that an intelligent agent could first learn a model of the world, and then use that internal model to plan its actions. The Dreamer series later demonstrated that many complex tasks could be learned by training agents inside an “imagined world.” At the same time, the development of video models such as Sora and Veo led researchers to another realization: A model capable of continuously generating video has already learned, at least implicitly, many of the rules governing the real world. As a result, these two research directions have gradually begun to converge. But Video Is Not Yet a World This is where the distinction is often misunderstood. For a world model to support meaningful real-time interaction, it must solve several critical problems. Most video models today are essentially answering one question: What should the next frame look like? A true world model needs to answer much more: What happens if I take one step forward? If I walk behind a building and then return, will the building still be there? If I suddenly change the camera angle, will the entire space remain consistent? If I enter a command such as: “Summon a dragon.” Will the world respond immediately? In other words, a world model must do more than generate content. It must understand space. It must understand time. It must understand causality. And it must understand interaction. Moving from watching to participating is where the real difficulty of world models begins. World Models Are Entering the Interactive Era One of the latest attempts in this direction is Alaya World, recently open-sourced by Alaya World, or Alaya Lab. Instead of generating a fixed video clip, it generates a world that users can explore in real time. Users can begin with text, an image, or a video, enter the generated scene, move freely through it, and introduce new prompts at any moment during generation. The world responds immediately. According to the publicly released information, Alaya World provides: Real-time streaming generation at 720p and 24 FPS Stable continuous exploration for more than one minute The ability to switch prompts and trigger skills or events during generation Model weights and inference code released under the Apache 2.0 License Training code and datasets planned for future release What makes these capabilities important is not simply the technical specifications. It is that the generated “world” can now support continuous interaction. The official demo shows that users can genuinely control, transform, and explore the generated environment. AI Is Evolving From a Tool Into an Environment Over the past few years, most discussions around AI have focused on content generation. Generating text. Generating images. Generating videos. But world models raise a fundamentally different question: Can AI generate an environment that people can inhabit, explore, and continuously evolve? If the answer is yes, the impact will extend far beyond video generation. Game development, robotics training, embodied intelligence, digital twins, virtual production, and many other fields could be transformed by the development of world models. World models are still at a very early stage. Yet from Craik’s proposal of an internal mental model more than eighty years ago to the emergence of today’s interactive world-generation systems, a clear evolutionary path is beginning to take shape. Perhaps what AI is ultimately learning has never been limited to images, videos, or language. Perhaps it is learning the world itself. References GitHub: Technical Report:

AI Is Moving Beyond “Generating Videos” — Toward “Generating Worlds” Over the past two years, AI video models have advanced at an astonishing pace. From Runway and Pika to Sora and Veo, AI-generated videos have become increasingly realistic and more consistent with the physical laws of the real world. Many people believe the next objective is simply to generate videos that are longer, sharper, and more lifelike. But if we take a step back, we can see that the real transformation is not happening in video itself. It is happening in world models. What Is a World Model? In 1943, psychologist Kenneth Craik proposed an idea that would influence artificial intelligence research for decades. He argued that the human brain does not merely react to the outside world. Instead, it maintains an internal model of how the world works. Because we have this internal model, we can predict the outcome of an action before we actually take it. Before crossing a road, we estimate whether a car will pass by. Before catching a ball, we predict its trajectory. These abilities come from continuously simulating the world in our minds, rather than relying entirely on trial and error. This idea later became known by a more formal term: World Model. A world model does not describe a single image or a fixed video clip. It is an internal representation capable of continuously simulating the rules and dynamics of the real world. Why Is AI Research Turning Toward World Models? Because predicting “what comes next” is becoming increasingly central to how AI systems work. Language models predict the next token. Image models predict the next step in the denoising process. Video models predict the next frame. A world model, however, attempts to predict something broader: What should the world look like in the next moment? In 2018, David Ha and Jürgen Schmidhuber proposed in their paper World Models that an intelligent agent could first learn a model of the world, and then use that internal model to plan its actions. The Dreamer series later demonstrated that many complex tasks could be learned by training agents inside an “imagined world.” At the same time, the development of video models such as Sora and Veo led researchers to another realization: A model capable of continuously generating video has already learned, at least implicitly, many of the rules governing the real world. As a result, these two research directions have gradually begun to converge. But Video Is Not Yet a World This is where the distinction is often misunderstood. For a world model to support meaningful real-time interaction, it must solve several critical problems. Most video models today are essentially answering one question: What should the next frame look like? A true world model needs to answer much more: What happens if I take one step forward? If I walk behind a building and then return, will the building still be there? If I suddenly change the camera angle, will the entire space remain consistent? If I enter a command such as: “Summon a dragon.” Will the world respond immediately? In other words, a world model must do more than generate content. It must understand space. It must understand time. It must understand causality. And it must understand interaction. Moving from watching to participating is where the real difficulty of world models begins. World Models Are Entering the Interactive Era One of the latest attempts in this direction is Alaya World, recently open-sourced by Alaya World, or Alaya Lab. Instead of generating a fixed video clip, it generates a world that users can explore in real time. Users can begin with text, an image, or a video, enter the generated scene, move freely through it, and introduce new prompts at any moment during generation. The world responds immediately. According to the publicly released information, Alaya World provides: Real-time streaming generation at 720p and 24 FPS Stable continuous exploration for more than one minute The ability to switch prompts and trigger skills or events during generation Model weights and inference code released under the Apache 2.0 License Training code and datasets planned for future release What makes these capabilities important is not simply the technical specifications. It is that the generated “world” can now support continuous interaction. The official demo shows that users can genuinely control, transform, and explore the generated environment. AI Is Evolving From a Tool Into an Environment Over the past few years, most discussions around AI have focused on content generation. Generating text. Generating images. Generating videos. But world models raise a fundamentally different question: Can AI generate an environment that people can inhabit, explore, and continuously evolve? If the answer is yes, the impact will extend far beyond video generation. Game development, robotics training, embodied intelligence, digital twins, virtual production, and many other fields could be transformed by the development of world models. World models are still at a very early stage. Yet from Craik’s proposal of an internal mental model more than eighty years ago to the emergence of today’s interactive world-generation systems, a clear evolutionary path is beginning to take shape. Perhaps what AI is ultimately learning has never been limited to images, videos, or language. Perhaps it is learning the world itself. References GitHub: Technical Report:

雪踏乌云

113,347 views • 2 months ago

🚨BREAKING: flyingtulip.com is preparing a public token sale with a built-in refund mechanism. One of the most prolific defi builders, Andre Cronje, famous for his work with yearn and Fantom/Sonic, has returned with the launch of Flying Tulip. The innovative part of Flying Tulip? The deposit and refund mechanism from the token sale: All deposits are deployed into Aave yield, with the team taking only the generated returns. Depositors receive FT tokens and can claim a full refund at 10c at any time via on-chain contracts. A programmatic price floor auto-buys if the token dips below 10c. Will this new model be the future of token raises? Read more below 👇

🚨BREAKING: flyingtulip.com is preparing a public token sale with a built-in refund mechanism. One of the most prolific defi builders, Andre Cronje, famous for his work with yearn and Fantom/Sonic, has returned with the launch of Flying Tulip. The innovative part of Flying Tulip? The deposit and refund mechanism from the token sale: All deposits are deployed into Aave yield, with the team taking only the generated returns. Depositors receive FT tokens and can claim a full refund at 10c at any time via on-chain contracts. A programmatic price floor auto-buys if the token dips below 10c. Will this new model be the future of token raises? Read more below 👇

BSCN

11,507 views • 7 months ago

I MADE MY AI AGENT 10X FASTER WITHOUT CHANGING THE MODEL not a smarter model, not a bigger context window, not another clever prompt the same kind of AI that designs vaccines for viruses we have not even met yet was spending two minutes opening the wrong files just to hand me a brief from three months ago the problem was never capability, it was the scaffolding that piled up around my agent by accident, folder by folder an agent does not think in your categories, it searches from scratch every single time, and your tidy human folders are a maze to it the fix was almost stupidly small, one index file at the root of each big folder and a few numbers in front of the folder names slowest task dropped from 2 minutes to 26 seconds, fastest ones hit 10, zero model changes capability is cheap when the scaffolding around it is broken the article breaks down the whole system in 15 minutes ↓

I MADE MY AI AGENT 10X FASTER WITHOUT CHANGING THE MODEL not a smarter model, not a bigger context window, not another clever prompt the same kind of AI that designs vaccines for viruses we have not even met yet was spending two minutes opening the wrong files just to hand me a brief from three months ago the problem was never capability, it was the scaffolding that piled up around my agent by accident, folder by folder an agent does not think in your categories, it searches from scratch every single time, and your tidy human folders are a maze to it the fix was almost stupidly small, one index file at the root of each big folder and a few numbers in front of the folder names slowest task dropped from 2 minutes to 26 seconds, fastest ones hit 10, zero model changes capability is cheap when the scaffolding around it is broken the article breaks down the whole system in 15 minutes ↓

shmidt

36,479 views • 2 months ago

This a a bit of fun- I set up a chatGPT shortcut as the action button for my Apple Watch and now I’m using it 10x more. It’s what Siri could have been / could be! This, custom instructions, and comments on churn/growth for AI apps in my latest blog post- (And this time w the right video 😂)

This a a bit of fun- I set up a chatGPT shortcut as the action button for my Apple Watch and now I’m using it 10x more. It’s what Siri could have been / could be! This, custom instructions, and comments on churn/growth for AI apps in my latest blog post- (And this time w the right video 😂)

andrew chen

29,048 views • 3 years ago

Chop the gradients ✂️! We found that truncating decoder gradients in latent video diffusion to a fixed window allows us to finetune on videos with pixel-wise perceptual losses without running out of memory. Pixel losses have been essential for image generation and reconstruction, but until now, they haven't scaled to long-duration, high-resolution video diffusion due to recursive activation accumulation in causal decoders, leading to OOM during training 💥📉. Project: Video diffusion models can do a lot more 🚀 when you can backprop the decoder! Post-process neural rendered scenes, super-resolve videos, harmonize lighting in controlled synthetic driving scenes, and inpaint videos — all in a single step ⚡ with a quick finetune from a standard diffusion model.

Chop the gradients ✂️! We found that truncating decoder gradients in latent video diffusion to a fixed window allows us to finetune on videos with pixel-wise perceptual losses without running out of memory. Pixel losses have been essential for image generation and reconstruction, but until now, they haven't scaled to long-duration, high-resolution video diffusion due to recursive activation accumulation in causal decoders, leading to OOM during training 💥📉. Project: Video diffusion models can do a lot more 🚀 when you can backprop the decoder! Post-process neural rendered scenes, super-resolve videos, harmonize lighting in controlled synthetic driving scenes, and inpaint videos — all in a single step ⚡ with a quick finetune from a standard diffusion model.

Felix Heide

28,467 views • 5 months ago

Tonight I left my laptop turned on, and woke up in the morning with a $7,000 profit. While I was sleeping, my cloud second brain was analyzing trends and running funnels on autopilot Famous investor Naval Ravikant once brilliantly noted: AI brains are meant for coming up with ideas, not clogging bio-servers with gigabytes of other people's garbage The main point Point Claude at a local folder and feed it all the accumulated digital junk files, articles, reports, notes. The artificial intelligence absorbs this chaos itself, builds ironclad logic, and turns a pile of files into a single, living knowledge base. The system works on the principle of compound interest: the more data you feed it, the more accurately it predicts your next steps and generates ready-made solutions Step-by-step process: Choose a base, either Notion or Obsidian, as long as the cloud is accessible from any device 24/7 Implement the CODE method by saving all the chaos to your inbox, sorting it out, extracting the essence, and monetizing your knowledge Connect AI agents by setting up automation via API so the AI sorts data and generates content on its own while you rest A more detailed guide is described in the article. I highly recommend adding this to your bookmarks

Tonight I left my laptop turned on, and woke up in the morning with a $7,000 profit. While I was sleeping, my cloud second brain was analyzing trends and running funnels on autopilot Famous investor Naval Ravikant once brilliantly noted: AI brains are meant for coming up with ideas, not clogging bio-servers with gigabytes of other people's garbage The main point Point Claude at a local folder and feed it all the accumulated digital junk files, articles, reports, notes. The artificial intelligence absorbs this chaos itself, builds ironclad logic, and turns a pile of files into a single, living knowledge base. The system works on the principle of compound interest: the more data you feed it, the more accurately it predicts your next steps and generates ready-made solutions Step-by-step process: Choose a base, either Notion or Obsidian, as long as the cloud is accessible from any device 24/7 Implement the CODE method by saving all the chaos to your inbox, sorting it out, extracting the essence, and monetizing your knowledge Connect AI agents by setting up automation via API so the AI sorts data and generates content on its own while you rest A more detailed guide is described in the article. I highly recommend adding this to your bookmarks

Bober_smart

12,381 views • 25 days ago

🇨🇳🇺🇸 China's military may be learning from America's AI, without building it from scratch Why spend billions training a frontier AI model when you can let someone else do the expensive part? Chinese military-linked researchers have repeatedly used outputs from OpenAI and Anthropic models to train smaller domestic AI systems for surveillance, cyber operations, drone targeting, battlefield decision-making, and software analysis. The trick is called distillation: instead of copying the model itself, researchers use its answers, and more importantly, its reasoning, to teach a smaller model. The result? AI that isn't as powerful as the original, but is cheap, runs on local hardware, and can be deployed on drones, military networks, and tactical systems without relying on U.S technology. The AI arms race isn't just about who builds the smartest model anymore; it's about who can squeeze the most military value out of everyone else's. Source: Reuters / Writer: Ian

🇨🇳🇺🇸 China's military may be learning from America's AI, without building it from scratch Why spend billions training a frontier AI model when you can let someone else do the expensive part? Chinese military-linked researchers have repeatedly used outputs from OpenAI and Anthropic models to train smaller domestic AI systems for surveillance, cyber operations, drone targeting, battlefield decision-making, and software analysis. The trick is called distillation: instead of copying the model itself, researchers use its answers, and more importantly, its reasoning, to teach a smaller model. The result? AI that isn't as powerful as the original, but is cheap, runs on local hardware, and can be deployed on drones, military networks, and tactical systems without relying on U.S technology. The AI arms race isn't just about who builds the smartest model anymore; it's about who can squeeze the most military value out of everyone else's. Source: Reuters / Writer: Ian

Mario Nawfal

47,775 views • 1 month ago