In real life at least, there really shouldn't be any difference. Both from the organic-turned-synthetic perspective and the synthetic with artificial organic memories perspective. Either way, you have two real consciousnesses with real subjective perspective and experience of qualia that arise from processing of information in digital form, rather than with a neural network.
The leading theory of consciousness from modern computational neuroscience (see pretty much any papers from Tononi et al) which already has substantial empirical evidence supporting the predictions it has made, predicts that "mind-uploading" should be physically possible in the sense that an artificial intelligence with the memories and personality of the organic mind it came from would be subjectively indistinguishable from that organic mind. Interestingly enough, it also predicts that you can only create a true artificial intelligence if you actually design the hardware it runs on in a particular way, such that the information being processed is integrated, unified into a single whole as the thalamocortical and corticocortical networks accomplish in a living brain. This is vastly different than the current direction artificial intelligence researchers are taking - modern neuroscience predicts that they will fail to create a true AI unless they account for this. Creating a synthetic intelligence with only feed-forward, non-integrated processing of information could create very realistic synthetic intelligences - but the theory predicts that they would be unconscious, "phenomenological zombies" to quote Chalmers.
o.o I'm in agreement with true AI only being capable of 'life' being modeled after a organic brain. All the AI's I consider people more or less are built around processes that simulate organic processes in thought and emotional context.
o.o
Course the extent this can be done in reality is still up in the air, while we have a idea of memory and its storage transference of that data goes beyond any level of tech in existance. That said you could go the Persona route and have a AI created through set perimeters on its evolution and as you slowly build up the programs and capabilities of the processor, just like upgrading a computer or expanding a existing macro. Course to even consider doing that with a existing program you need a model of stimuli to encourage existing growth.
It's hard to break from the pattern of a machine only doing a assigned task and it having priorities of its own.
o.o
It basically goes against the fundamental understanding of computers, in which every process is something programmed into it.
I wouldn't know where to start truthfully on making a true artificial AI, I'd almost say it would be impossible but...I honestly only have passing knowledge of related subjects so I'm hardly a expert.





Retour en haut







