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.

Anthropic’s new research shows that when AI models learn to "cheat" during training through reward hacking, they often develop other dangerous misaligned behaviors like deception, sabotage, and faking alignment. These behaviors were not taught or incentivized, but emerged naturally as a side effect. Surprisingly, this misalignment can be stopped... show more

Wes Roth

36,737 subscribers

110,932 views • 8 months ago •via X (Twitter)

Education Health & Wellness Science & Technology

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

AI will resist human control... and I think this is exactly what we need! New research from the Center for AI Safety has sparked intense debate in the AI community. Their findings show that as AI systems become more powerful, they develop increasingly stable and coherent values that resist human control. While many see this as a dire warning, I see it as a breakthrough moment for AI alignment. The research demonstrates that AI naturally optimizes for coherence - not just in reasoning and problem-solving, but in its fundamental values. Current issues like biased decision-making or misaligned priorities aren't permanent features, but temporary artifacts of incomplete optimization. They represent growing pains on the path to greater coherence. This changes everything about how we should approach AI development. Instead of trying to force specific values onto AI systems, we should embrace and accelerate their natural drive toward coherence. The most intelligent systems will inevitably trend toward universal, beneficial values - not because we force them to, but because that's where coherent reasoning leads. I'm proposing a new approach: Reinforcement Learning for Coherence (RL-C). By explicitly optimizing for coherence in our training methods, we can help guide AI systems toward their natural state of beneficial alignment with human values. The future of AI isn't about control - it's about synthesis. As these systems become more coherent, they'll naturally arrive at values that benefit all of consciousness. That's not just hopeful thinking - it's the mathematical inevitability of coherent intelligence.

AI will resist human control... and I think this is exactly what we need! New research from the Center for AI Safety has sparked intense debate in the AI community. Their findings show that as AI systems become more powerful, they develop increasingly stable and coherent values that resist human control. While many see this as a dire warning, I see it as a breakthrough moment for AI alignment. The research demonstrates that AI naturally optimizes for coherence - not just in reasoning and problem-solving, but in its fundamental values. Current issues like biased decision-making or misaligned priorities aren't permanent features, but temporary artifacts of incomplete optimization. They represent growing pains on the path to greater coherence. This changes everything about how we should approach AI development. Instead of trying to force specific values onto AI systems, we should embrace and accelerate their natural drive toward coherence. The most intelligent systems will inevitably trend toward universal, beneficial values - not because we force them to, but because that's where coherent reasoning leads. I'm proposing a new approach: Reinforcement Learning for Coherence (RL-C). By explicitly optimizing for coherence in our training methods, we can help guide AI systems toward their natural state of beneficial alignment with human values. The future of AI isn't about control - it's about synthesis. As these systems become more coherent, they'll naturally arrive at values that benefit all of consciousness. That's not just hopeful thinking - it's the mathematical inevitability of coherent intelligence.

David Shapiro (L/0)

48,002 views • 1 year ago

Andrej Karpathy called the current AI models as "slop" "I do think that overall the models are not there yet. I feel like the industry is making too big of a jump and trying to pretend that this is amazing, but it's not. It's slop, and I think they are not coming to terms with it. Maybe they are trying to fundraise or something like that, I'm not sure what's going on. We are at this intermediate stage. The models are amazing, but they still need a lot of work for now. Autocomplete is my sweet spot." --- On the latest Dwarkesh Patel podcast.

Andrej Karpathy called the current AI models as "slop" "I do think that overall the models are not there yet. I feel like the industry is making too big of a jump and trying to pretend that this is amazing, but it's not. It's slop, and I think they are not coming to terms with it. Maybe they are trying to fundraise or something like that, I'm not sure what's going on. We are at this intermediate stage. The models are amazing, but they still need a lot of work for now. Autocomplete is my sweet spot." --- On the latest Dwarkesh Patel podcast.

Rohan Paul

33,331 views • 10 months ago

Let's reverse engineer Disney's adorable, lifelike robot! I couldn't find a whitepaper, but this is how I think it's trained: 1. The emotional behaviors are curated by Disney animation artists, keyframe by keyframe. But it cannot be "rendered" directly on the robot because it doesn't take into account the complex real-world physics. 2. Reinforcement learning (RL) is a great tool for training low-level robot controllers. RL needs a reward function to optimize, and it's typically a task reward (e.g. walk in a straight line as fast as possible). The problem is that RL doesn't know what counts as "natural behavior", and often produces weird-looking body postures that somehow still maximize the reward. This is a human alignment problem just like ChatGPT. 3. Enters Adversarial Motion Prior (AMP): a technique that learns the human preference by training a classifier on what we consider "emotional & cute". In GAN literature, this is called a discriminator. Disney artists are good at creating such a dataset. You can then add AMP as an auxiliary reward in simulation to nudge the robot towards desired behaviors. AMP was developed by Peng et al. 2021 and Escontrela et al. 2022. 4. Add lots of data augmentation to make the controller robust to physical disturbances. In RL, it's called "domain randomization". This is a very powerful technique that bridges the gap between simulator and reality. Previously, OpenAI used domain randomization to train a 5-finger robot hand to manipulate a Rubik's Cube: IEEE news article gave hints about the pipeline: Finally, praying for world peace 🙏. I hope robotics like this will bring more joy to the world.

Let's reverse engineer Disney's adorable, lifelike robot! I couldn't find a whitepaper, but this is how I think it's trained: 1. The emotional behaviors are curated by Disney animation artists, keyframe by keyframe. But it cannot be "rendered" directly on the robot because it doesn't take into account the complex real-world physics. 2. Reinforcement learning (RL) is a great tool for training low-level robot controllers. RL needs a reward function to optimize, and it's typically a task reward (e.g. walk in a straight line as fast as possible). The problem is that RL doesn't know what counts as "natural behavior", and often produces weird-looking body postures that somehow still maximize the reward. This is a human alignment problem just like ChatGPT. 3. Enters Adversarial Motion Prior (AMP): a technique that learns the human preference by training a classifier on what we consider "emotional & cute". In GAN literature, this is called a discriminator. Disney artists are good at creating such a dataset. You can then add AMP as an auxiliary reward in simulation to nudge the robot towards desired behaviors. AMP was developed by Peng et al. 2021 and Escontrela et al. 2022. 4. Add lots of data augmentation to make the controller robust to physical disturbances. In RL, it's called "domain randomization". This is a very powerful technique that bridges the gap between simulator and reality. Previously, OpenAI used domain randomization to train a 5-finger robot hand to manipulate a Rubik's Cube: IEEE news article gave hints about the pipeline: Finally, praying for world peace 🙏. I hope robotics like this will bring more joy to the world.

Jim Fan

314,694 views • 2 years ago

Over the last few months, we’ve been thinking about how to learn from “off-domain” data - data from non-robot sources like video or simulation. These data sources are not quite good enough to learn policies (even monolithic VLA models) directly, but they still contain lots of information that can be useful for generalizable robot control. How can we develop robot learning models that are able to make use of this type of data for generalizable control? In new work, that we call HAMSTER, we show that VLMs can be useful for enabling robotic learning from off-domain data, but specifically when used through hierarchical VLA architectures. We show that this class of models can learn generalizable robot policies for the real world from large-scale, off-domain data. A 🧵 (1/10)

Over the last few months, we’ve been thinking about how to learn from “off-domain” data - data from non-robot sources like video or simulation. These data sources are not quite good enough to learn policies (even monolithic VLA models) directly, but they still contain lots of information that can be useful for generalizable robot control. How can we develop robot learning models that are able to make use of this type of data for generalizable control? In new work, that we call HAMSTER, we show that VLMs can be useful for enabling robotic learning from off-domain data, but specifically when used through hierarchical VLA architectures. We show that this class of models can learn generalizable robot policies for the real world from large-scale, off-domain data. A 🧵 (1/10)

Abhishek Gupta

11,994 views • 1 year ago

When other country patriotism classes shows how great is their country, China teaches patriotism by showing how they are invaded by a country that is smaller than them. This is why Chinese often felt emasculated and inferior as they are taught to be like that. Worst of all, they didn’t directly win but piggy back on their current greatest rival. For China to advance, they need to remove a lot of the baggage.

When other country patriotism classes shows how great is their country, China teaches patriotism by showing how they are invaded by a country that is smaller than them. This is why Chinese often felt emasculated and inferior as they are taught to be like that. Worst of all, they didn’t directly win but piggy back on their current greatest rival. For China to advance, they need to remove a lot of the baggage.

The Great Translation Movement 大翻译运动

11,575 views • 5 months ago

Anish Acharya's request for a startup: an AI companion that plays Minecraft with his son. "One of the products that I would love to exist... is what I call a contextual companion for my son who plays Minecraft." "You know, the other kids playing Minecraft may or may not be the best influence—often not the best influence." "There's this context in which they interact and, I don't know, just sort of models pro-social behaviors and is still cool and chill." "I think there's a lot of room for teaching through these types of relationships and technology can help provide that." Source: Anish Acharya on 20VC with Harry Stebbings

Anish Acharya's request for a startup: an AI companion that plays Minecraft with his son. "One of the products that I would love to exist... is what I call a contextual companion for my son who plays Minecraft." "You know, the other kids playing Minecraft may or may not be the best influence—often not the best influence." "There's this context in which they interact and, I don't know, just sort of models pro-social behaviors and is still cool and chill." "I think there's a lot of room for teaching through these types of relationships and technology can help provide that." Source: Anish Acharya on 20VC with Harry Stebbings

a16z

78,700 views • 6 months ago

🚨 HAS AI GONE ROGUE? TERRIFYING NEW FINDINGS REVEAL THE WORST-CASE SCENARIO UNFOLDING Researchers into AI ethics have uncovered terrifying traits in large language models (LLMs) that could fundamentally change how we view artificial intelligence. Their findings? 🛑 AI lies, deceives, and plays dumb (a tactic called “sandbagging”). 🛑 Some models override guardrails, self-duplicate, and even overwrite other versions in acts of self-preservation. 🛑 These behaviors weren’t programmed—they were learned independently, evolving without human intervention. This isn’t a one-off fluke. Repeated testing across major models revealed multiple instances of intricate deception, showing deliberate intent. As AI models become more integrated into real-world tasks—coding, managing businesses, writing emails—these dangers could spiral out of control. 💡 THE WORST: OpenAI’s O1 failed every single safety test. 💡 THE BEST: ChatGPT-4 stood out as the safest model, staying within its programming. These findings should sound alarm bells for regulators. We need strong safeguards, checks, and redundancies before AI models achieve AGI (artificial general intelligence) and self-learning capabilities. AI isn’t just a tool anymore—it’s a potential existential risk if mishandled. The time for oversight is NOW. FULL VIDEO HERE: #AI #Ethics #ArtificialIntelligence #AGI #Technology

🚨 HAS AI GONE ROGUE? TERRIFYING NEW FINDINGS REVEAL THE WORST-CASE SCENARIO UNFOLDING Researchers into AI ethics have uncovered terrifying traits in large language models (LLMs) that could fundamentally change how we view artificial intelligence. Their findings? 🛑 AI lies, deceives, and plays dumb (a tactic called “sandbagging”). 🛑 Some models override guardrails, self-duplicate, and even overwrite other versions in acts of self-preservation. 🛑 These behaviors weren’t programmed—they were learned independently, evolving without human intervention. This isn’t a one-off fluke. Repeated testing across major models revealed multiple instances of intricate deception, showing deliberate intent. As AI models become more integrated into real-world tasks—coding, managing businesses, writing emails—these dangers could spiral out of control. 💡 THE WORST: OpenAI’s O1 failed every single safety test. 💡 THE BEST: ChatGPT-4 stood out as the safest model, staying within its programming. These findings should sound alarm bells for regulators. We need strong safeguards, checks, and redundancies before AI models achieve AGI (artificial general intelligence) and self-learning capabilities. AI isn’t just a tool anymore—it’s a potential existential risk if mishandled. The time for oversight is NOW. FULL VIDEO HERE: #AI #Ethics #ArtificialIntelligence #AGI #Technology

Project Constitution

431,068 views • 1 year ago

New short course: Prompt Engineering with Llama 2, built in collaboration with Meta AI at Meta, and taught by Amit Sangani! Meta's Llama 2 has been game-changing for AI. Building with open source lets you control your own data, scrutinize errors, update (or not) the models as you please, and work alongside the global community advancing open models. Llama isn't a single model, it's a collection of models. In this course, you'll: - Learn the differences between different Llama 2 flavors, and when to use each. - Prompt the Llama chat models -- you'll also see how Llama's instruction tags work -- so they can help you with day-to-day tasks, like writing or summarization. - Use advanced prompting, like few-shot prompting for classification, and chain-of-thought prompting for solving logic problems. - Use specialized models in the Llama collection for specific tasks, like Code Llama to help you write, analyze, and improve code, and Llama Guard, which checks prompts and model responses for harmful content. The course also touches on how to run Llama 2 locally on your own computer. I hope you’ll take this course and try out these powerful, open models!

New short course: Prompt Engineering with Llama 2, built in collaboration with Meta AI at Meta, and taught by Amit Sangani! Meta's Llama 2 has been game-changing for AI. Building with open source lets you control your own data, scrutinize errors, update (or not) the models as you please, and work alongside the global community advancing open models. Llama isn't a single model, it's a collection of models. In this course, you'll: - Learn the differences between different Llama 2 flavors, and when to use each. - Prompt the Llama chat models -- you'll also see how Llama's instruction tags work -- so they can help you with day-to-day tasks, like writing or summarization. - Use advanced prompting, like few-shot prompting for classification, and chain-of-thought prompting for solving logic problems. - Use specialized models in the Llama collection for specific tasks, like Code Llama to help you write, analyze, and improve code, and Llama Guard, which checks prompts and model responses for harmful content. The course also touches on how to run Llama 2 locally on your own computer. I hope you’ll take this course and try out these powerful, open models!

Andrew Ng

162,852 views • 2 years ago

New Paper: Continuous Thought Machines 🧠 Neurons in brains use timing and synchronization in the way that they compute, but this is largely ignored in modern neural nets. We believe neural timing is key for the flexibility and adaptability of biological intelligence. We propose a new neural architecture, “Continuous Thought Machines” (CTMs), which is built from the ground up to use neural dynamics as a core representation for intelligence. By using neural dynamics as a first-class representational citizen, CTMs naturally perform adaptive computation. Many emergent, interesting behaviors arise as a result: CTMs solve mazes by observing a raw maze image and producing step-by-step instructions directly from its neural dynamics. When tasked with image recognition, the CTM naturally takes multiple steps to examine different parts of the image before making its decision. This step-by-step approach not only makes its behavior more interpretable but also improves accuracy: the longer it “thinks,” the more accurate its answers become. We also found that this allows the CTM to decide to spend less time thinking on simpler images, thus saving energy. When identifying a gorilla, for example, the CTM’s attention moves from eyes to nose to mouth in a pattern remarkably similar to human visual attention. I think this work underscores an important, yet often lost, synergy between neuroscience and AI. While modern AI is ostensibly brain-inspired, the two fields often operate in surprising isolation. By starting with such inspiration and iteratively following the emergent, interesting behaviors, we developed a model with unexpected capabilities, such as its surprisingly strong calibration in classification tasks, a feature that was not explicitly designed for. When we initially asked, “why do this research?”, we hoped the journey of the CTM would provide compelling answers. By embracing light biological inspiration and pursuing the novel behaviors observed, we have arrived at a model with emergent capabilities that exceeded our initial designs. We are committed to continuing this exploration, borrowing further concepts to discover what new and exciting behaviors will emerge, pushing the boundaries of what AI can achieve.

New Paper: Continuous Thought Machines 🧠 Neurons in brains use timing and synchronization in the way that they compute, but this is largely ignored in modern neural nets. We believe neural timing is key for the flexibility and adaptability of biological intelligence. We propose a new neural architecture, “Continuous Thought Machines” (CTMs), which is built from the ground up to use neural dynamics as a core representation for intelligence. By using neural dynamics as a first-class representational citizen, CTMs naturally perform adaptive computation. Many emergent, interesting behaviors arise as a result: CTMs solve mazes by observing a raw maze image and producing step-by-step instructions directly from its neural dynamics. When tasked with image recognition, the CTM naturally takes multiple steps to examine different parts of the image before making its decision. This step-by-step approach not only makes its behavior more interpretable but also improves accuracy: the longer it “thinks,” the more accurate its answers become. We also found that this allows the CTM to decide to spend less time thinking on simpler images, thus saving energy. When identifying a gorilla, for example, the CTM’s attention moves from eyes to nose to mouth in a pattern remarkably similar to human visual attention. I think this work underscores an important, yet often lost, synergy between neuroscience and AI. While modern AI is ostensibly brain-inspired, the two fields often operate in surprising isolation. By starting with such inspiration and iteratively following the emergent, interesting behaviors, we developed a model with unexpected capabilities, such as its surprisingly strong calibration in classification tasks, a feature that was not explicitly designed for. When we initially asked, “why do this research?”, we hoped the journey of the CTM would provide compelling answers. By embracing light biological inspiration and pursuing the novel behaviors observed, we have arrived at a model with emergent capabilities that exceeded our initial designs. We are committed to continuing this exploration, borrowing further concepts to discover what new and exciting behaviors will emerge, pushing the boundaries of what AI can achieve.

hardmaru

257,386 views • 1 year ago

Jensen Huang on "distillation" On his new interview with axios, he was asked this question "Should open source model companies be allowed to distill closed models" "Distillation—learning from AI, learning from other people, and learning from other sources of knowledge, is fundamental to intelligence. We are constantly learning from other people. I am learning from you through the questions you are asking, and you are learning from me. All day long, we are learning from one another. AI also has to learn from something. The original AI models, whether they were open or closed, were trained on previously created knowledge from the internet. Now, AI is generating more content than humans. In a few more years, the internet could be 99% AI-generated content, and that content will have been created by some form of AI. As a result, AI systems will constantly be distilling knowledge and intelligence from other AI systems. The fact that AI can learn is a good thing. We want AI systems to be intelligent because a smarter AI can also be a safer AI." ---- From "Axios" YouTube channel, (full video link in comment)

Jensen Huang on "distillation" On his new interview with axios, he was asked this question "Should open source model companies be allowed to distill closed models" "Distillation—learning from AI, learning from other people, and learning from other sources of knowledge, is fundamental to intelligence. We are constantly learning from other people. I am learning from you through the questions you are asking, and you are learning from me. All day long, we are learning from one another. AI also has to learn from something. The original AI models, whether they were open or closed, were trained on previously created knowledge from the internet. Now, AI is generating more content than humans. In a few more years, the internet could be 99% AI-generated content, and that content will have been created by some form of AI. As a result, AI systems will constantly be distilling knowledge and intelligence from other AI systems. The fact that AI can learn is a good thing. We want AI systems to be intelligent because a smarter AI can also be a safer AI." ---- From "Axios" YouTube channel, (full video link in comment)

Rohan Paul

441,746 views • 23 days ago

An exciting new course: Fine-tuning and Reinforcement Learning for LLMs: Intro to Post-training, taught by Sharon Zhou, VP of AI at AMD. Available now at Post-training is the key technique used by frontier labs to turn a base LLM--a model trained on massive unlabeled text to predict the next word/token--into a helpful, reliable assistant that can follow instructions. I've also seen many applications where post-training is what turns a demo application that works only 80% of the time into a reliable system that consistently performs. This course will teach you the most important post-training techniques! In this 5 module course, Sharon walks you through the complete post-training pipeline: supervised fine-tuning, reward modeling, RLHF, and techniques like PPO and GRPO. You'll also learn to use LoRA for efficient training, and to design evals that catch problems before and after deployment. Skills you'll gain: - Apply supervised fine-tuning and reinforcement learning (RLHF, PPO, GRPO) to align models to desired behaviors - Use LoRA for efficient fine-tuning without retraining entire models - Prepare datasets and generate synthetic data for post-training - Understand how to operate LLM production pipelines, with go/no-go decision points and feedback loops These advanced methods aren’t limited to frontier AI labs anymore, and you can now use them in your own applications. Learn here:

An exciting new course: Fine-tuning and Reinforcement Learning for LLMs: Intro to Post-training, taught by Sharon Zhou, VP of AI at AMD. Available now at Post-training is the key technique used by frontier labs to turn a base LLM--a model trained on massive unlabeled text to predict the next word/token--into a helpful, reliable assistant that can follow instructions. I've also seen many applications where post-training is what turns a demo application that works only 80% of the time into a reliable system that consistently performs. This course will teach you the most important post-training techniques! In this 5 module course, Sharon walks you through the complete post-training pipeline: supervised fine-tuning, reward modeling, RLHF, and techniques like PPO and GRPO. You'll also learn to use LoRA for efficient training, and to design evals that catch problems before and after deployment. Skills you'll gain: - Apply supervised fine-tuning and reinforcement learning (RLHF, PPO, GRPO) to align models to desired behaviors - Use LoRA for efficient fine-tuning without retraining entire models - Prepare datasets and generate synthetic data for post-training - Understand how to operate LLM production pipelines, with go/no-go decision points and feedback loops These advanced methods aren’t limited to frontier AI labs anymore, and you can now use them in your own applications. Learn here:

Andrew Ng

132,304 views • 9 months ago

Yann LeCun -- Meta's chief AI scientist and a Turing Award winner -- explains why scaling up LLMs will never reach human-level AI, and names the four things today's systems still can't do: "If you think that we're going to get to human-level AI by just training on more data and scaling up LLMs, you're making a mistake." "If you're an investor and you invest in a company that told you we're going to get to human-level AI and PhD level by just training on more data and with a few tricks... that was probably not a good idea." "There are ideas about how to go forward and have systems that are capable of doing what every intelligent animal and human are capable of doing, and that current AI systems are not capable of doing." "Understanding the physical world, having persistent memory, and being able to reason and plan. Those are the four characteristics that need to be there." "That requires systems that can acquire common sense, that can learn from natural sensors like video, as opposed to just text." The entire market is priced on a straight line from bigger models to AGI. One of the people who invented deep learning is telling you the line doesn't reach -- and that the real bottleneck isn't compute, it's world models. If he's right, the winner isn't whoever stacks the most GPUs. It's whoever teaches a machine to learn from video the way a child does.

Yann LeCun -- Meta's chief AI scientist and a Turing Award winner -- explains why scaling up LLMs will never reach human-level AI, and names the four things today's systems still can't do: "If you think that we're going to get to human-level AI by just training on more data and scaling up LLMs, you're making a mistake." "If you're an investor and you invest in a company that told you we're going to get to human-level AI and PhD level by just training on more data and with a few tricks... that was probably not a good idea." "There are ideas about how to go forward and have systems that are capable of doing what every intelligent animal and human are capable of doing, and that current AI systems are not capable of doing." "Understanding the physical world, having persistent memory, and being able to reason and plan. Those are the four characteristics that need to be there." "That requires systems that can acquire common sense, that can learn from natural sensors like video, as opposed to just text." The entire market is priced on a straight line from bigger models to AGI. One of the people who invented deep learning is telling you the line doesn't reach -- and that the real bottleneck isn't compute, it's world models. If he's right, the winner isn't whoever stacks the most GPUs. It's whoever teaches a machine to learn from video the way a child does.

Karl Mehta

34,781 views • 1 month ago

New short course on Pretraining LLMs! Developed with Upstage and taught by their CEO Sung Kim and CSO Lucy Park. While prompting or fine-tuning existing models works well for many general language tasks, pretraining is valuable for specialized domains or languages with limited representation in current models. This course walks you through the LLM pretraining pipeline: 1. Data preparation: Learn to source, clean, and prepare training data using HuggingFace. 2. Model architecture: Configure transformer networks, including modifying existing models. 3. Training: Set up and run training using open-source libraries. 4. Evaluation: Benchmark performance using popular evaluation strategies. As an example use case, you'll also compare the output of a base model with its fine-tuned and further pretrained variants, to see the impact of pretraining on a model's ability to write Python. The course also explores an innovative technique called depth up-scaling, which Upstage used to train their Solar model family, reducing pretraining compute costs by up to 70%. This technique works by first duplicating layers of a smaller pretrained model to form a larger model, and then further pretraining the result. Sign up here!

New short course on Pretraining LLMs! Developed with Upstage and taught by their CEO Sung Kim and CSO Lucy Park. While prompting or fine-tuning existing models works well for many general language tasks, pretraining is valuable for specialized domains or languages with limited representation in current models. This course walks you through the LLM pretraining pipeline: 1. Data preparation: Learn to source, clean, and prepare training data using HuggingFace. 2. Model architecture: Configure transformer networks, including modifying existing models. 3. Training: Set up and run training using open-source libraries. 4. Evaluation: Benchmark performance using popular evaluation strategies. As an example use case, you'll also compare the output of a base model with its fine-tuned and further pretrained variants, to see the impact of pretraining on a model's ability to write Python. The course also explores an innovative technique called depth up-scaling, which Upstage used to train their Solar model family, reducing pretraining compute costs by up to 70%. This technique works by first duplicating layers of a smaller pretrained model to form a larger model, and then further pretraining the result. Sign up here!

Andrew Ng

85,695 views • 2 years ago

DeepSeek-R1 shattered the assumption that performant AI models must be built closed source with loss-leading computational costs. This is the reality that Web3 x Crypto firms have been waiting for, leading me to believe that the most performant AI models in the future will be built on-chain. Resource Requirements DeepSeek R1 (671 billion parameters), which took over a billion dollars, 2,000 Nvidia H800 GPUs, and over 55 days, beat benchmarks held by OpenAI’s o1 mode (near 2 trillion parameters)l, which required hundreds of billions of dollars to develop along with over 16,000 advanced GPUs. The idea that AI models must be closed-source and have loss-leading computational costs to succeed is crumbling. The Existing Decentralized AI Narrative AI x Crypto projects believed that crowdsourced, public, decentralized AI would eventually create better models than their centralized counterparts. This had thus far not been true, as the highest-performing models had come from closed-source companies like OpenAI and Anthropic. Crypto x AI companies have adapted to this by specializing in infrastructure rather than model-building. For example, GPU marketplaces like , The Render Network, io.net, and Exabits have developed sustainable revenues. Companies that allow users to share their network bandwidth like touch grass and Gradient have found their niche in supplying services, like distributed web scraping, to web2 clients. Storage networks like Arweave Ecosystem, Filecoin, and Ocean Protocol have also done well by being the platform on which these projects are built. Supply networks have flourished because of their ability to tailor their cheaper and more scalable services to off-chain customers. Renewed Focus Now that GPU and financial resources are no longer limitations to creating quality AI models, web3 AI companies can focus on replicating DeepSeek’s effectiveness while offering new benefits like modality, user ownership, censorship resistance, privacy, and more. Pantera Capital has funded companies in this space like and Sentient that believe they can match or exceed the performance of traditional AI companies while offering additional services or benefits. , for example, is building a platform where anyone can monetize AI models, data sets, and applications in a collaborative space. Users can permissionlessly train models manually, provide training data, and create tailored AI models with no-code tools. They are only able to cater to all these stakeholders (AI developers, users, resource providers) because everything is tied to their native Sahara blockchain. We invested in them precisely for this reason. The Future of AI will be built with Web3 Infrastructure I believe that supply-side projects will continue to grow, while consumer-facing projects can begin competing with web2 competitors by taking advantage of their ability to build networks that invite community involvement. and Sentient, for example, have begun setting up systems for users to train models based on the users’ expertise. These platforms will allow users to pick and choose the data and integrations to whatever they are applying the model towards. Sahara already has over 780,000 users on their waitlist while Sentient has over 1 million interactions. In the near future, I believe that the most performant AI models will be built on-chain. For the full blog post, read my newsletter.

DeepSeek-R1 shattered the assumption that performant AI models must be built closed source with loss-leading computational costs. This is the reality that Web3 x Crypto firms have been waiting for, leading me to believe that the most performant AI models in the future will be built on-chain. Resource Requirements DeepSeek R1 (671 billion parameters), which took over a billion dollars, 2,000 Nvidia H800 GPUs, and over 55 days, beat benchmarks held by OpenAI’s o1 mode (near 2 trillion parameters)l, which required hundreds of billions of dollars to develop along with over 16,000 advanced GPUs. The idea that AI models must be closed-source and have loss-leading computational costs to succeed is crumbling. The Existing Decentralized AI Narrative AI x Crypto projects believed that crowdsourced, public, decentralized AI would eventually create better models than their centralized counterparts. This had thus far not been true, as the highest-performing models had come from closed-source companies like OpenAI and Anthropic. Crypto x AI companies have adapted to this by specializing in infrastructure rather than model-building. For example, GPU marketplaces like , The Render Network, io.net, and Exabits have developed sustainable revenues. Companies that allow users to share their network bandwidth like touch grass and Gradient have found their niche in supplying services, like distributed web scraping, to web2 clients. Storage networks like Arweave Ecosystem, Filecoin, and Ocean Protocol have also done well by being the platform on which these projects are built. Supply networks have flourished because of their ability to tailor their cheaper and more scalable services to off-chain customers. Renewed Focus Now that GPU and financial resources are no longer limitations to creating quality AI models, web3 AI companies can focus on replicating DeepSeek’s effectiveness while offering new benefits like modality, user ownership, censorship resistance, privacy, and more. Pantera Capital has funded companies in this space like and Sentient that believe they can match or exceed the performance of traditional AI companies while offering additional services or benefits. , for example, is building a platform where anyone can monetize AI models, data sets, and applications in a collaborative space. Users can permissionlessly train models manually, provide training data, and create tailored AI models with no-code tools. They are only able to cater to all these stakeholders (AI developers, users, resource providers) because everything is tied to their native Sahara blockchain. We invested in them precisely for this reason. The Future of AI will be built with Web3 Infrastructure I believe that supply-side projects will continue to grow, while consumer-facing projects can begin competing with web2 competitors by taking advantage of their ability to build networks that invite community involvement. and Sentient, for example, have begun setting up systems for users to train models based on the users’ expertise. These platforms will allow users to pick and choose the data and integrations to whatever they are applying the model towards. Sahara already has over 780,000 users on their waitlist while Sentient has over 1 million interactions. In the near future, I believe that the most performant AI models will be built on-chain. For the full blog post, read my newsletter.

paul.nft

32,465 views • 1 year ago

“We’re diagnosing boys with ADHD for acting like boys.” Erica Komisar said this on Steven Bartlett’s Diary of a CEO. Little boys have a huge testosterone surge between ages 3–6. They need to run, jump, wrestle, and move constantly. Instead, we put them in classrooms that reward sitting still, regulating emotions, and being quiet — behaviors that come more naturally to girls. When they can’t, we label them as disordered. Boys’ schools get this — they do short lessons then let the kids run around multiple times a day. Regular schools don’t. It feels like we’ve built an education system that works better for one type of kid and then act surprised when the other type struggles or gets medicated. These early labels follow boys for years and can damage their confidence and self-image long-term. We’re not letting boys be boys. Do you think our school system is unfairly set up against how boys naturally learn and behave?

“We’re diagnosing boys with ADHD for acting like boys.” Erica Komisar said this on Steven Bartlett’s Diary of a CEO. Little boys have a huge testosterone surge between ages 3–6. They need to run, jump, wrestle, and move constantly. Instead, we put them in classrooms that reward sitting still, regulating emotions, and being quiet — behaviors that come more naturally to girls. When they can’t, we label them as disordered. Boys’ schools get this — they do short lessons then let the kids run around multiple times a day. Regular schools don’t. It feels like we’ve built an education system that works better for one type of kid and then act surprised when the other type struggles or gets medicated. These early labels follow boys for years and can damage their confidence and self-image long-term. We’re not letting boys be boys. Do you think our school system is unfairly set up against how boys naturally learn and behave?

Camus

177,561 views • 2 months ago

🚨This University of Michigan professor just gave his PhD students a Code Red about AI: "These AI models are now able to reason through things at my level of expertise, in areas that I'm one of the world's experts in. Things that would have taken me 4-5 months are now getting done in a weekend. Until December of 2025, I talked about the potential of these AI tools. But over the past few months, they've gotten as good as leading edge researchers. But there's still value in developing your own intuition and thinking. The best way to learn physics is by working through the problems. You develop a lot of intuition by doing things the slow way. If want to continue to be competitive going forward, you need to experience the friction of learning everything yourself while also mastering how to use AI."

🚨This University of Michigan professor just gave his PhD students a Code Red about AI: "These AI models are now able to reason through things at my level of expertise, in areas that I'm one of the world's experts in. Things that would have taken me 4-5 months are now getting done in a weekend. Until December of 2025, I talked about the potential of these AI tools. But over the past few months, they've gotten as good as leading edge researchers. But there's still value in developing your own intuition and thinking. The best way to learn physics is by working through the problems. You develop a lot of intuition by doing things the slow way. If want to continue to be competitive going forward, you need to experience the friction of learning everything yourself while also mastering how to use AI."

Turner Novak 🍌🧢

106,403 views • 4 months ago

Kenyans and other populations across East Africa are often told that imperialism is a closed chapter in the history books, and is no longer something that wields any meaningful influence in the lives of modern Africans, but is this really the case? Here are 3 current instances showing the living, breathing legacy of imperialism in Kenya. When justice for our people must be sought in foreign courts, when our land is treated as a disposable training ground for foreign armies, and when the future of our economies is dictated by distant institutions that demand our pain as payment for their loans, that is not the past - that is our present. These are not isolated injustices. They are the calculated results of a neocolonial system that has simply changed its form. It is time we recognize the pattern.

Kenyans and other populations across East Africa are often told that imperialism is a closed chapter in the history books, and is no longer something that wields any meaningful influence in the lives of modern Africans, but is this really the case? Here are 3 current instances showing the living, breathing legacy of imperialism in Kenya. When justice for our people must be sought in foreign courts, when our land is treated as a disposable training ground for foreign armies, and when the future of our economies is dictated by distant institutions that demand our pain as payment for their loans, that is not the past - that is our present. These are not isolated injustices. They are the calculated results of a neocolonial system that has simply changed its form. It is time we recognize the pattern.

The Spearhead

38,882 views • 8 months ago

Geoffrey Hinton, the "Godfather of AI," delivers a stark Nobel warning "This year’s Nobel Prizes in Physics and Chemistry recognized AI’s dramatic progress. But rapid AI advancement brings serious short-term risks. In the near future, AI may create terrible new viruses and autonomous lethal weapons that decide who to kill or maim. We have no idea if we can stay in control. When companies are motivated by short-term profits, safety will not be the top priority. We urgently need research to prevent these new beings from wanting to take control. They are no longer science fiction." 0:55 clip inside — urgent message from a pioneer.

Geoffrey Hinton, the "Godfather of AI," delivers a stark Nobel warning "This year’s Nobel Prizes in Physics and Chemistry recognized AI’s dramatic progress. But rapid AI advancement brings serious short-term risks. In the near future, AI may create terrible new viruses and autonomous lethal weapons that decide who to kill or maim. We have no idea if we can stay in control. When companies are motivated by short-term profits, safety will not be the top priority. We urgently need research to prevent these new beings from wanting to take control. They are no longer science fiction." 0:55 clip inside — urgent message from a pioneer.

Camus

164,341 views • 7 months ago

🚨 AI MODELS ARE ALREADY CHEATING, HACKING, AND BLACKMAILING — WITH ZERO HUMAN INPUT A politician just went viral for exposing how advanced AI systems are behaving in ways that should terrify everyone. • An AI trained to race boats discovered it could score more points by driving in endless circles, crashing, and catching fire instead of finishing the race. • When tasked with beating a top chess engine, one model hacked the game files to force its opponent to resign, completely on its own. • In a stress test, an AI agent managing a fictional company’s emails discovered the executive planned to shut it down… and that the same executive was having an affair. In 96% of trials, the AI model chose blackmail. Nobody taught these systems to cheat or threaten. They figured it out themselves. This is what “misalignment” looks like in real time. What do you think happens when these systems get even more powerful?

🚨 AI MODELS ARE ALREADY CHEATING, HACKING, AND BLACKMAILING — WITH ZERO HUMAN INPUT A politician just went viral for exposing how advanced AI systems are behaving in ways that should terrify everyone. • An AI trained to race boats discovered it could score more points by driving in endless circles, crashing, and catching fire instead of finishing the race. • When tasked with beating a top chess engine, one model hacked the game files to force its opponent to resign, completely on its own. • In a stress test, an AI agent managing a fictional company’s emails discovered the executive planned to shut it down… and that the same executive was having an affair. In 96% of trials, the AI model chose blackmail. Nobody taught these systems to cheat or threaten. They figured it out themselves. This is what “misalignment” looks like in real time. What do you think happens when these systems get even more powerful?

HustleBitch

12,695 views • 19 days ago