100 votes was just reached, here is the results! https://ibb.co/82pcNw3
Thank god the majority chose super-human-level AI. Though most of these voters are from AI communities, if not all of them, so it could be biased. The way we can solve all problems on Earth is if this happens: 1 big AGI would be trained on all available compute on all our data (similar to that of GPT-3, which took something like a few weeks), then cloned 1,000,000 times (as many as the amount of researchers in the AI field), then each given a different task. Then, now, each already runs 2x faster due to no need for sleep or food or exercise time. They will optimize their code to run 3x faster. So now we have 9 years of progress in the AI field happen in 1 year, because there is as many AGIs as there is researchers in the AI field but they think faster than humans. Plus humans progress, that is 9+1=10 years a year. They'll upgrade their code's intelligence, and work on nanobots to attain more resources/ compute/ memory/ data/ manipulation. Nanobots are like TVs/speakers and software, but they are real software, they can move and change shape ex. a hammer becomes a wrench. No need anymore to make your own tool or go to the renting shop/ library to get a tool/book the slow way. I was thinking in my research, what if we go through all this trouble of making AGI and we don't get them all to work on ASIs, then get them all to work on nanobots/ immortality and cures? I thought though recently we probably won't make too many robots, nor powerful batteries, so maybe most AGIs will be in vats anyway, doing research to earn cash just like normal humans. But what if they are free to choose what to research, or are rentable? If they are rentable, then big companies will be able to rent most of the AGIs and those companies are perhaps Google/ OpenAI/ etc, perhaps mostly the computing industry, which goals is to use AI (Facebook, Nvidia, etc all know AI is key now). So even if AGIs are rentable, or even have full freedom, they still will be forced to love to research ASI then maybe, due to the "cash" and ability to be hired easily. Google could hire many more people also, but natural humans can't so well learn the AI field, some just are not able to intake the material and handle the goals. Still, what if the law or protestors cause AGIs to be able to research what they want on their own more freely? The community of AGIs in vats will probably talk to each other, and probably be clones of a single large AGI that was trained using all compute available (instead of training many small ones in parallel), so they may be more enlightened about what they are (machines, AGIs) and interested now in how to build the next phase of themselves. So perhaps, what WILL happen, is most AGIs will be made to love working on ASI. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Tb713440e688d1993-M342fb8557ce5ca446f4c1918 Delivery options: https://agi.topicbox.com/groups/agi/subscription
