President of Dopamine 🍉♺ @socialdopamine.bsky.social · 21/09/2026There's a threshold we have to cross before we have to fear AI. It's the recursive self-improvement threshold. Supposedly, once this is crossed, AI can improve itself beyond our ability to control it. But here's why I'm doubtful. 100 0
President of Dopamine 🍉♺ @socialdopamine.bsky.social · 21/09/2026AI is going to hit diminishing returns, not exponential improvement. A brain in a vacuum feeding on information can't jump 1000 years ahead of us on science because it won't have 1000 years worth of lab experiments, physical observations, etc. 100 0
