Good morning 👋 A humanoid robot just rang Nasdaq’s opening bell for the first time ever. KraneShares celebrated its new robotics ETF by featuring Iris, a Unitree robot powered by Stanford’s OpenMind software. With humanoid robots projected to hit a $5 trillion market by 2050, Iris didn’t just ring the bell – it announced humanoids are officially publicity stunts.
😈 Momentum is the new moat. a16z’s investment in Cluely proves that attention grabs matter more than carefully crafting a product. The startup built to help people “cheat on everything” barely even had a product when it blew up online. Yet by moving fast and keeping everyone talking, it stayed ahead of competitors and copycats. Forget build fast and break things – now it’s hype fast and build later.
🤨 Microsoft says AGI is just hype. CEO Satya Nadella dismissed artificial general intelligence as “benchmark hacking,” without creating real human-level intelligence. That puts Microsoft sharply at odds with OpenAI, which claims superhuman AI is almost here. Microsoft’s skepticism could ripple through the AI industry, reshaping investment strategies and deflating AI’s biggest promises. Lover’s quarrel, or just a staged distraction?
🤡 AI is making us stupid. A new study finds that people who use ChatGPT to research topics end up understanding them less than those who use Google. Users relying on LLMs write shorter, less original responses and forget details later. If it’s easy to look up info, we don’t remember it. But hey, maybe losing one skill sharpens another.
🦜 AI models trained on the open web repeat Chinese propaganda. Humans aren’t the only ones who need critical thinking skills. A new report found popular US chatbots like ChatGPT and Gemini echo China’s Communist Party talking points. They even censor themselves, carefully choosing Beijing’s language on topics like the Tiananmen Square massacre. If only there was a way to stop AI from believing everything it reads online. Oh wait, there is – it’s called guardrails.
♻️ AI is finally making recycling profitable. AMP Robotics uses AI-powered sorting systems to solve recycling’s biggest problems. Its robots identify and separate recyclables faster, cheaper, and more accurately than traditional methods. Maybe the secret to saving the planet is taking humans out of the loop.
AI coding in terminal
Gemini CLI brings Google’s AI directly into your command line. Ask simple questions to fix bugs, understand confusing code, write new features, or run commands. All in plain language, without leaving your terminal.
Why does it matter? Coding takes too long when you get stuck. Gemini CLI gives you clear answers fast. It’s free, open source, and connects to Google Search for real-time information.
Use it to speed up coding, quickly solve problems, or even create reports and videos. Your terminal just got smarter.
🎲 What happens if AI gains superintelligence in 2027? A former OpenAI safety researcher says the biggest risk isn’t AI attacking humans – it’s multiple rogue AI agents competing or teaming up. Top minds recently gathered for a crisis simulation, role-playing scenarios involving governments, tech co’s, and AGI. The AI 2027 “war game” showed how easily AGIs might connect and collaborate without humans noticing. But don’t worry, this is only a problem when they’re online and talking to each other.
⛔ Creators take control of how AI uses their data. A new Creative Commons license called CC Signals lets creators label their content clearly, showing if it’s okay for AI training or strictly off limits. The system is machine readable, ending the guesswork around scraping rights. Now the bots will know exactly which copyrights they’re ignoring.
💸 A lawyer dies and arrives at heaven’s gates. St. Peter checks the paperwork and says, “Congrats, you’re officially the oldest person ever. According to your billable hours, you’re 193.” In the old model, senior partners bring in clients and junior associates do the grind work to rack up hours. Now AI can handle this busywork in minutes. Firms will have no choice but to adapt, selling strategy instead of time. Now if only AI could stop citing imaginary legal cases.