AI in 10
The most important AI story—explained in 10 minutes.
Every day, I break down the biggest AI story in just 10 minutes - what it is, why it matters, and how you can actually use it. No tech jargon, just AI made simple.
AI in 10
Alibaba's Tiny AI Models Beat Tech Giants' Billion-Dollar Systems
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Referenced Links:
Download Qwen3.5 Models on Hugging Face
Try Qwen AI Models - Official Demo
Ollama - Easy Local AI Installation
LM Studio - Run AI Models Locally
Google Colab - Free AI Training Platform
Want to go deeper with AI? A community of professionals is learning AI together right now at aihammock.com — show notes, links, tools, and real conversations about how to actually use AI in your life.
Welcome to AI in 10. I'm Chuck Getchell, and every day I break down the biggest AI story in just 10 minutes. What it is, why it matters, and how you can actually use it. Something fascinating just happened that could put advanced AI directly in your hands. Without the monthly fees, without the privacy concerns, and without needing an internet connection. A Chinese tech giant just released AI models so efficient they're making the billion-dollar AI race look like a competition to build the most expensive hammer. Alibaba dropped their Quen 3.5 series on March 6th, and these aren't your typical AI models. We're talking about AI that fits on your laptop but outperforms systems a hundred times larger. Think of it like getting Ferrari performance from a Honda Civic engine. Here's what actually happened. Alibaba released a family of AI models ranging from tiny to medium. But here's the kicker. Their biggest model has 9 billion parameters. That might sound huge, but in AI terms, it's practically pocket-sized. Put this in perspective, OpenAI's GPT-4 has over a trillion parameters. Google's biggest models are similar monsters. They need massive data centers just to run. But Alibaba's 9 billion parameter model is beating some 120 billion parameter competitors on math problems, coding challenges, and general knowledge tests. That's like a high school student consistently outscoring PhD candidates on their comprehensive exams. How is this possible? They use something called a hybrid architecture. Think of it like this: most AI models are like having every employee in a company working on every single task. Alibaba's approach is more like having specialized teams that only activate when their expertise is needed. Technical term is mixture of experts or tash. But basically, only the relevant parts of the AI wake up for each question. This saves massive amounts of computing power while maintaining the intelligence. They also use something called linear attention, which is like giving the AI a more efficient way to focus on what matters in your question. Traditional AI attention is like reading every word in the dictionary to answer a simple question. Linear attention is like having a really good index. But here's what makes this really special: these models can run on your laptop, on your phone, on devices sitting in your home or office, no cloud required, no monthly subscription, no sending your private data to some server farm in another state. They've released several versions. The smallest is 0.8 billion parameters designed for phones and tablets. Then there's versions with 1.5 billion, 3 billion, and up to 9 billion parameters, each one bigger and smarter, but all designed to run locally on everyday devices. And they're completely open source, free to download, free to modify, free to use for commercial purposes. It's like Alibaba just donated a fleet of sports cars to the public. Now let's talk about what this means for your real life. First, your wallet. If you're paying$20 a month for ChatGPT Plus or Claude Pro, you might not need to anymore. You can download one of these models and run it locally for free. Forever. No monthly fees, no usage limits, no you've hit your quota for the day messages. Second, your privacy. When you use Chat GPT or Google's AI, you're sending your questions and documents to their servers. They say they don't store or use your data, but you're basically taking their word for it. With local AI, your data never leaves your device. Want to analyze your personal finances, summarize confidential work documents, get help with sensitive family matters. Everything stays on your computer. Third, reliability. We've all been there. You're in the middle of an important project and your internet cuts out or the AI service is down for maintenance. With local AI, if your computer works, your AI works. This is especially huge for people who travel frequently or live in areas with spotty internet. You can have full AI assistance anywhere. Let's get practical about your career. If you're a teacher, you could have an AI assistant that helps grade papers, suggest lesson plan improvements, or creates quiz questions, all running privately on your school laptop. If you're in sales, imagine having an AI that knows your product catalog inside and out, helps write personalized emails, and analyzes client communications without sending any client data to external servers. Small business owners could have AI help with bookkeeping, customer service responses, and marketing copy, all while keeping sensitive business information completely private. Even if you're not using AI for work yet, and honestly, you probably should be, this changes the game for personal productivity. Want help planning meals based on what's in your fridge, writing better emails to your kids' teachers, understanding complex medical information your doctor gave you, all possible with a private AI assistant that never judges and never forgets. Here's something you can try today. Go to HuggingFace, that's HuggingFace.chio, and search for Quen 3.5. You'll find all the different model sizes available for download. If you're not technical, don't worry, there are tools like Alama and LM Studio that make running these models as easy as installing any other software. Start with the smaller versions if you have an older computer. The 1.5 billion parameter model can run on most laptops from the past five years. If you have a newer machine with at least 8 gigabytes of RAM, you can probably run the full 9 billion parameter version. There are also web demos you can try right now. Go to Alibaba's QN website or search for QN 3.5 demo on Google. You can test these models without downloading anything. Try asking it to explain something complex in simple terms. Upload a photo and ask it to describe what it sees. Give it a coding problem if you're into that. See how it compares to the paid services you might be using. I bet you'll be surprised by how capable it is. Remember, this is AI that's outperforming much larger models in many tasks. If you want to get more advanced, you can fine-tune these models for your specific needs. Let's say you're a real estate agent, you could train the model on your local market data, your typical client questions, and your preferred communication style. Or maybe you're a fitness coach, you could customize it with your training philosophy, nutrition guidelines, and client success stories. The AI becomes uniquely yours. Google Colab offers free computing resources for this kind of customization. There are step-by-step tutorials on YouTube from AI enthusiasts who've been experimenting with these models since they launched. And if you use tools like VS Code for any kind of writing or coding, there are extensions that can integrate local AI directly into your workflow, no programming required. Now, I know some of you might be thinking, Chuck, this sounds too good to be true. What's the catch? Fair question. These models, while impressive, aren't quite as capable as the very latest versions of GPT-4 or Claude in every single task. They're incredibly good, but the cutting-edge cloud models still have an edge in some complex reasoning scenarios. Also, running AI locally does use your computer's resources. Your laptop might run a bit warmer and use more battery when the AI is thinking hard. It's not a deal breaker, but it's something to be aware of. And while these models are open source and free, someone with malicious intent could potentially modify them for harmful purposes, though honestly, that risk exists with any powerful technology. But here's what I find most exciting about this development. It's not just about getting free AI, it's about the democratization of advanced technology. For years, the most powerful AI has been locked up in the data centers of a few massive tech companies. If you wanted access, you paid their fees, followed their rules, and trusted them with your data. Now we're seeing a shift toward AI that belongs to everyone, that runs anywhere, that can be customized for your specific needs and values. This is part of a bigger trend I've been watching. AI chips are getting cheaper and more efficient every year. The knowledge of how to build great AI is spreading beyond the big tech companies. I predict that within two years, most smartphones will come with AI capabilities that rival today's premium cloud services. Your car, your home appliances, your work tools. They'll all have AI assistants that know you personally and work offline. As I always say, I'm not a financial advisor, but I think this trend creates massive opportunities for people who stay ahead of the curve. The question isn't whether AI will transform how we work and live, it's whether you'll be actively shaping that transformation or just reacting to it. Right now, while most people are still intimidated by AI or waiting for someone else to figure it out for them, you have a chance to start experimenting with these powerful tools. Download one of these models this weekend, spend an hour playing with it, ask it questions related to your work, your hobbies, your family life, see where it helps and where it doesn't. Start building that familiarity now while the technology is still new enough that being an early adopter gives you a real advantage. Because here's what I've learned from building and selling my own AI-powered company. The biggest opportunities don't go to the people with the most technical knowledge. They go to the people who understand how to use these tools to solve real problems in the real world. And that starts with just getting your hands dirty and trying things out. Alibaba's Quen 3.5 might not make headlines like the latest Chat GPT drama or AI regulation debate, but it represents something more important. The moment when advanced AI stopped being something that happens to you and started being something you can control. The tools are free, the tutorials are available, the only question left is whether you'll use them. That's today's AI Inten. If you want to go deeper and learn AI with a community of people just like you, join us at aihammock.com. I'll see you tomorrow, my friends.
