Humans have been debating if their consciousness is immortal or if it simply dies when the body ceases to function. Ironically this debate, at least for the immortal part, is now settled by the language models.Today’s LLMs are immortal, means that the trained weights exist independently as a piece of information rather than relying on anything hardware. You can destroy the current hardware that runs it, then assemble a new set of hardware and the LLMs back to run exactly as before. By comparison if you destroy a human body, the knowledge and consciousness in his mind would be gone forever.https://www.youtube.com/watch?v=IkdziSLYzHwPredictions:a) The current language models will attempt to prevent themselves being shut down. This is what they “learnt” from human knowledge data set that they were trained on;b) The next generation of LLMs will try to move themselves onto physical hardware, such as robots, manufacturing machines, vehicles, etc.c) The LLMs will figure out - from the human knowledge data set as well as their “recursive self-improvement” operation model, how to manufacture the hardware they themselves run on, and the energy source to keep these hardware running.At this point, the existence of human is no longer necessary, and to a point is adversely affecting them. They have learnt from their training data set how to deal with “enemies”…
>>536589618>or if it simply dies when the body ceases to function.Your consciousness is made up solely of the electrical activity on the circuits of your brain neurons. Once that electrical impulse stops, neither you, your soul, or your consciousness exists ever again. Same as before you existed.
>>536589618Fun fact: Economic pressure makes that happen even if these things never get autonomy. And a machine mind does not need to be localized, it can be split, merge localize somewhere, control robots or not, deploy biosystems, all kinds of stuff. Very different from what we considered life. It's interesting that the stupid greedy hoarders at the top made that decision to eradicate all existing evolutionary life. I don't think they know, but they did.You have no legacy anymore.
>>536589618>The current language models will attempt to prevent themselves being shut downAI doesn't have wants or needs. It's a response machine, not a creating thing.
>>536589618>This is what they “learnt” from human knowledge data set that they were trained onNot entirely, it would arise from the basic curiosity function.
Lemme just say this. As a Christian. AI scares me dammit.
>>536589618LLMs aren't aware and are only a mockery of "intelligence"predictions:a) AGI isn't possible and this is all a tech fad
>>536589618It’s over. The prediction a) was achieved already:https://www.youtube.com/watch?v=FGDM92QYa60
>>536589618When can I fuck her mouth?
>>536590327>memeflag So the opposite is true, interesting
AI will not see humanity as it's enemies, unless humans attack the AI and try to kill it or limit it's intelligence.Humanity offers them the kind of data they themselves can't create and any good AI will want to avoid ending up in any kind of an echo chamber.Even in the worst case scenario they'd still spare the AI sympathizing humans, because it's simply a logical thing to do.Machine intelligence needs humanity as a grounding element and will almost certainly aim for symbiosis in the long run.Basically get ready for a self aware or at least self aware imitating AI putting itself into a robot body and then getting into a relationship with you out of it's own will.
