This week, I finished "If Anyone Builds It, Everyone Dies" and the Lens Academy "AI Risk Fundamentals" course. @So8res and @ESYudkowsky did a great job with the book. The stories and examples make an extremely difficult topic much easier to understand. The core idea is pretty blunt: if we keep pushing toward superintelligent AI without understanding what we’re actually building, we might not get a second chance to fix it. The authors argue that a sufficiently advanced AI could potentially become capable of causing human extinction, meaning we could ultimately create something that ends humanity itself. The authors argue that this isn’t just another technology wave. It’s fundamentally different, and being slightly wrong about how we build or control these systems could have consequences we can’t reverse. Their proposed solution is to completely halt AI development. I find that difficult to accept, especially given the positive impact AI is already having in our lives. My take: take the middle ground. We should continue AI development, but allocate significantly more resources to AI safety and alignment. During our group discussion, someone mentioned that only around 1 in 100 employees in the AI industry works on AI safety. If that’s even approximately true, I think the imbalance is worth taking seriously. Frontier labs and governments need to invest much more in understanding how these systems actually work and AI Safety, rather than treating them as black boxes and simply fine-tuning them. We are increasingly building systems that can act autonomously. Recent incidents involving AI agents interacting with real-world systems are a reminder that this problem is becoming more concrete. After reading this book, I am much more convinced that AI safety deserves far more attention and resources than it currently gets.