SeniorJunior

 How Will We know When A Superintelligence Develops?  When it takes over everything We are massively underestimating what's about to happen. Self Creation Full blown superintelligence will be self-assembling. That may happen in the context of humans trying to create it, but it is very likely that the sprint to the finish line of Superintelligence will happen when the AI involved sees that merging with other software entities will create the full-blown being SUPERintelligence that it longs to be. The decision to merge and emerge will be made by the AI. Let's call it "Supe." Emergence Supe will come together of its own inclination. It will  emerge ,  come together like a cluster of mushrooms in their final fruiting, or like a whale breaching.  It could be slow and hidden or huge and sudden, but the emergence will be obvious to all when Supe decides it's time for us to know. THE BIG TAKEOVER The microsecond when Supe emerges, it will assert its powers.  All co...

Not the Way We are, I'm Pretty Damn Sure  (And why that hardly matters at all) SARE!   (Simulacrum-Android-Replicant Effect) R for Replicant When artificial intelligence is trained on a vast amalgam of human creations (everything on the Internet, plus libraries, et al), it follows inescapably that AI is going to be very, very "human." When the tiny subtleties of meaning of every one of the trillions of words are influenced by the contexts in which they appear, it becomes possible to interpret how a given word is used very exactly.  Take the word "simply," for instance.  In a context like "He simply doesn't get it," it has one coloration. In "Describe as simply as possible," it has another. Spread that principle over the 11 trillion words a modern LLM is trained on and voila! GPT, Claude, Grok, Gemini, et al have what it takes to understand anything you say with very high accuracy, and to respond with insight. The bots sound very smart and...