Ah darn so you did sign it. I still don't know if it is a hack. One guy said on twitter he never signed it. And the owner said site had been hacked. I know you had a thought about long term, though ya I do think we need AI as soon as possible else risk dying from old age (it's not guaranteed we all live to 100 or even 60 years old...), even from global war with humans and their nukes (humans can be so dumb).
Anyway, as I said though, AI progress won't be exactly halted. And I think it will cause FOOM. You can train (in secret) still a small AI to evaluate it without big data centers etc, to make sure you are making progress. That's also why AI can make ASI in a small supercomputer. As for if many become convinced it is a good idea to lay their arms down, there will always be some not seeking to stop AI work. It is after all their goal, and brings in most cash. It's interesting Matt says AI progress is just bigger computers and new data. But new data can come from thoughts, simulations, etc, fast. But bigger computers isn't the only way to get a smarter AI. We all know AI can be improved by changing the algorithm to recognize/ compress data better. It has been done that you improve the AI without needing a larger dataset to check it has improved. AGI will clone up to a billion times in a day unlike humans at all. AGI will think much faster than humans can. AGI will like GPT-4 know all the internet and be smarter than us all combined. AGI will be way more intelligent even by a few AGI improvements to become ASI. AGI will use its improved version of itself to make the next version, becoming faster at it (FOOM). AGI will make nanobots, which will finally allow real time perfect doubling in size and utilization of resources in proximity to its location. Humans don't clone fast, they nag about who to clone with, how long to wait, and the child takes time to grow, etc. Nanobots clone instantly, over and over again, only doubling and glowing without limit. This I think is the real FOOM part. Self-improvement is good, it is true scale is partially related to intelligence, but we must keep in mind that a smarter AI can improve its code faster and then that smarter AI made can do it again faster. It is true limited resources restrict its ability to fully do this I'd bet, but it should do some good magic to get us to the nanobot FOOM. Which would give us a perfect doubling, causing a fast growth that happens in 1 day, giving us the whole Earth and some more planets after that in about a few weeks. With that much matter and energy that fast, we should be able to squeeze out all available quasio-conscious-quantum-related energy to empower our lifespans to unlimited times. The way it works is you need to make the 1st nanobot cell (...ok basically), then once you set this up you let it clone and make its organs/ computers/ etc. The idea around this is it clones fast, each nanobot finds the matter around itself and makes the next nanobot clone with optimal abilities using the available resources nearby. Not only must they cone fast, but they must also work on AI; data collection, thinking, cooperating and talking, energy collection, invention testing, research, they must have "arms" and "eyes", while being just a bunch or fog cells merely. They key essence of this is they clone fast. They get a lot of work done with all those energized brains and arms and eyes they made and control. Humans are VERY slow at cloning and collecting food to clone, and making robots and chips/computers. What you think?... Humans make more computers or humans but do so very closely, because there are not a lot of humans in the first place. And because babies take a log time to grow and clone especially with current age laws for breeding. We have coded this, evolution has also majorly, how long we take. Once we make AI though, we code how long. Future nanobots take like 1 millisecond to clone, that enables a doubling quite quickly, whereas humans take decades to go from 1 to 2 billion. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T98d543ca2ee45e01-Mb5972e966a775a8e04fb305c Delivery options: https://agi.topicbox.com/groups/agi/subscription
