"But why would you think that AGI would not hallucinate?" Your "AGI" may hallucinate, because it is designed to feed on that incoherent second-hand natural-language data. Mine won't, it is designed to be integral and self-sufficient. It will believe what it sees, not what a bunch of nuts on the net say.
On Tue, Mar 5, 2019 at 2:52 PM Linas Vepstas <[email protected]> wrote: > > > On Tue, Mar 5, 2019 at 12:37 PM Matt Mahoney <[email protected]> > wrote: > >> Steve, good luck ending the political debate over climate change. But >> you have a few obstacles. >> >> 1. Overwhelming evidence does not end political debate. Just ask the >> creationists, anti-vaxxers, moon landing hoaxers, and 9/11 conspiracy >> theorists. The whole purpose of the flat earth society is not to >> convince you that the earth is flat, but to show that what you think >> are logical, sound, and obvious arguments supported by undisputed >> facts are actually useless. >> > > More precisely: computers have lowered the cost of publishing so low, that > anyone can publish: you no longer run a gauntlet of editors, printers and > proof-readers who tell you that your ideas are stupid. Facebook and twitter > have taken this to a new level. As a result, we can now all hear > each-others brains thinking, and, it turns out, they are incoherent, > contradictory and insane. Worse: this high-connectivity, high-bandwidth > (youtube) low-latency interconnect allows for the spread and amplification > of "memes" disconnected from any basis in reality. > > In short: social media has attached our individual, singular brains into a > big "global brain", and we are hearing that global brain think, and it is > hallucinating a lot of the time. Old-school social sciences have already > studied this: propaganda, (Hitler studied propaganda), cults, brainwashing, > Stockholm syndrome, and also plenty of less harmful things: everything from > pop music and Puerto-Rican low-rider automotive clubs. Uplifting things, > too: from scientists to medical doctors to humanitarian activists. > > If you want a preview of what a mildly super-human intelligence viz AGI > might think, then the hallucinatory beliefs of various memetic tribes is a > good sampling. The memeplex of creationists, anti-vaxxers, moon landing > hoaxers, and 9/11 conspiracy > theorists is sufficiently self-consistent to be stable, and more: > sufficient to be invasive, and occupy the thought-space, the noosphere of > many human brains. It spreads. > > Perhaps you think that an AGI will be purely "rational" (whatever that > means!) and that no AGI could ever be a creationist anti-vaxxer > moon-landing denialist. But why would you think that AGI would not > hallucinate? > > Until very recently, Reality, i.e. the universe, entrained thinking minds > in such a way that you starve and die, if you cannot think clearly enough > to obtain food and procreate. Squirrels who neglect nuts die sooner, rather > than later. Our modern economy is sufficiently robust that you can > hallucinate all day long, or watch soap operas on TV, or do whatever it is > you do, and mostly not starve to death. Mostly; we seem to have a problem > with Amazon employees in LA and SF. But whatever. > > If you want a safe, non-existential-threatening AGI in the future, > attempting to understand and control the root causes of hallucinatory > thinking today is a good place to start. > > -- Linas > -- > cassette tapes - analog TV - film cameras - you > *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + delivery > options <https://agi.topicbox.com/groups/agi/subscription> Permalink > <https://agi.topicbox.com/groups/agi/Tbefabf50a1da4070-M7f718fff7015cfc48c5c572b> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Tbefabf50a1da4070-M4c500c48685c384665cd0508 Delivery options: https://agi.topicbox.com/groups/agi/subscription
