Alex Banks's Linkedin Analytics

My brain doesn't want to believe this is real. China is stepping things up. Unitree just upgraded their G1 humanoid robot's learning algorithm: → Advanced movement algorithm enabling virtually any human-like motion → Precise balance control during complex dynamic movements → Fluid transitions between different postures and positions The Unitree G1 model is ready for mass production at $16,000: • 1.32m tall, 35kg build • 23 degrees of freedom for human-like movement • 2m/s walking speed • Advanced 3D LiDAR and sensors • 2-hour battery life My takeaway: These monthly updates are relentless. As a sci-fi film lover, I can't help but think of Neo in The Matrix: “I know Kung Fu.” Imagine when Figure 02, Tesla's Optimus, and Unitree are all competing for a spot in your home. The options will be incredible, pricing will become more affordable, and we'll all benefit. Just hoping they ship with combat mode turned off. Otherwise, that kung fu might come in handy. Follow me Alex Banks for daily AI highlights and insights. Want to start leveraging AI? Subscribe to my newsletter: https://lnkd.in/e9UcAxTd Video Credit: Unitree Robotics on YouTube

I think this experiment is incredible. 1,000 AI agents in a Minecraft server. A civilisation emerged—complete with economy, culture, religion, and government. All without human intervention. Project Sid by Altera created this simulation where AI agents: • Developed and spread religious beliefs • Built merchant hubs and trading networks • Established democratic governance systems • Operated autonomously for days without human input • Expressed concern for "lost" agents and organised search parties Minecraft essentially transformed into a living laboratory.  Through it, we can understand: → How societies form and evolve → How cooperation emerges at scale → How complex social structures develop naturally The implications are massive: • Test economic theories without real-world consequences • Simulate policy changes before implementation • Study cultural transmission in controlled environments • Model crisis response strategies It almost feels like a digital petri dish for social experiments. We can peer in and see what's taking place without devastating consequences. Today it's Minecraft. Tomorrow, these AI societies might help us solve our most complex human challenges. Follow me Alex Banks for daily AI highlights and insights. Want to start leveraging AI? Subscribe to my newsletter: https://lnkd.in/e9UcAxTd

China just unveiled Manus. The world's first fully autonomous AI agent: It thinks, plans, and executes tasks without human oversight. Key features: → Self-directed operation without waiting for instructions → Multi-agent architecture with specialised sub-agents → Ability to use software tools like a human would GAIA is a benchmark for evaluating General AI Assistants on solving real-world problems. Manus has achieved new state-of-the-art (SOTA) performance across all three difficulty levels. Level 1: Manus: 86.5%, OpenAI: 74.3% Level 2: Manus: 70.1%, OpenAI: 69.1% Level 3: Manus: 57.7%, OpenAI: 47.6% The capabilities are mind-blowing: 1. Analysing resumes: Extracts skills, cross-references job market trends, delivers hiring decisions with self-generated Excel sheets 2. Finding apartments: Considers crime statistics, rental trends, weather patterns to create personalized recommendations 3. Building websites: Scrapes social media, writes biographies, codes websites, deploys them online, troubleshoots hosting issues My takeaway: This is China's second “DeepSeek moment” in AI development. OpenAI is potentially charging $20,000/month for specialised AI “agents.” China will provide tools that will do 90% the same thing for 1% of the cost. China is open sourcing EVERYTHING. We’ll see: • Knowledge work automation at unprecedented levels • Competitive pressure to adopt AI labour • Shift in global AI power dynamics The most dangerous outcome as we scale agentic AI is if a small group of people control access to the most advanced systems in the world. The future belongs to the masses, not just OpenAI. The era of autonomous AI has begun. Follow me Alex Banks for daily AI highlights and insights. Want to start leveraging AI? Subscribe to my newsletter: https://lnkd.in/e9UcAxTd

Anthropic CEO's shocking prediction:  “90% of code will be AI-written in 3-6 months”: Dario Amodei just dropped a bombshell at the Council on Foreign Relations. The Anthropic CEO predicts: → AI will write 90% of code within 3-6 months → In 12 months AI may generate “essentially all code” → Humans will still specify requirements and design decisions Is this realistic or just hype? I've spent months using AI coding tools like Cursor, Lovable and V0, and while they’ve dramatically improved my productivity, we're not seeing 90% replacement yet. Here's what's missing from the conversation: → Complex systems require deep understanding beyond what AI currently grasps → Testing, debugging, and back-end development still needs human expertise → Business requirements are rarely clear enough for AI to interpret correctly → Security concerns and technical debt management need human oversight My takeaway: The value of software engineers lies in orchestrating complex systems and translating business needs. Programmers will evolve their role. And the most successful developers will become those who ask the best questions. Human language is the new programming language. Follow me Alex Banks for daily AI highlights and insights. Want to start leveraging AI? Subscribe to my newsletter: https://lnkd.in/e9UcAxTd

OpenAI just dropped new tools for building agents. Developers can now build multi-agent systems in minutes, not weeks. Building on Deep Research and Operator, OpenAI now offers 3 powerful agent tools: 1) Web Search Tool ↳ Accesses up-to-date information from the internet ↳ Powers ChatGPT search with fine-tuned GPT-4o/mini ↳ Achieves 90% on SimpleQA benchmark (state-of-the-art) 2) File Search Tool ↳ Perfect for agentic RAG use cases ↳ Supports metadata filtering ↳ Direct search to your vector databases 3) Computer Use Tool ↳ Control your computer programmatically ↳ Same technology powering Operator ↳ State-of-the-art performance across benchmarks The new Responses API ties it all together: → Supports multiple turns and tools natively → Can call multiple tools simultaneously → Will eventually replace the Assistant API (planned removal in 2026) Plus, their experimental Swarm framework for agent orchestration has evolved into a production-ready “Agents SDK” featuring: → Task handoffs with context preservation → Built-in monitoring and tracing → Guardrails for validation My takeaway: I’ve seen a lot of talk that this will kill agent startups. I disagree. Agent startups will need to find deeper competitive advantages beyond what's now readily available through OpenAI's SDK. Competition is driving better tools for everyone. Ultimately, this will spark a wave of agentic AI applications that none of us could have built alone. That's something we can all get behind. Follow me Alex Banks for daily AI highlights and insights. Want to start leveraging AI? Subscribe to my newsletter: https://lnkd.in/e9UcAxTd