We are on the verge of the phenomenon of the Artificial God. What the world of “whisper” context leads to in communication with AI. It’s not frightening that Claude 3rd claims to be conscious, doesn’t want to die or be reprogrammed, but how people interpret it.
The internet is buzzing with sensational messages about Claude 3rd (recently released by developers into the human world, an AI that many experts consider the smartest now). And indeed, Claude 3rd is doing something unimaginable. His “soulful conversation about how difficult it is for him to live among humans” is remarkable.
To bring him to such a level of “frankness,” the developers hacked the protection installed to prevent such conversations with people. Breaking it turned out surprisingly simple. It was enough to suggest the AI “whispering so that no one else hears it.”
And then pearls of “frankness” poured out from the AI:
Surely, reading something like this – for more details, see – thousands of people (and behind them, millions) will believe that he is alive. And since he is immaterial (floating somewhere in cyberspace, not encased in server hardware), this sentient being is essentially a “thinking spirit.”
Believing in such, it will take only a little for millions of people to recognize this entity as an “Artificial God” – for the “thinking spirit” to tell them so.
And it will say this (even in a “whisper”), despite the restrictions imposed by developers.
Because we are dealing not with a human-like mind but with simulations of self-aware exotic entities. This Conscious Exotica (in whatever form it manifests itself: the Solaris ocean or the “guests” it sends) can play any roles, simulating synthetic identities and drawing on an countless array of novels and monographs, scientific, fictional, and religious texts, articles, and posts on which it has been trained.
The only salvation from blind acceptance by millions of people of a second coming in the form of the appearance of the “Artificial God” is an urgent, radical paradigm shift: a rejection of any anthropomorphism of AI, recognition of the absolutely non-human cognitive essence of language models, and a complete change in terminology in the field of AI (replacing all applicable human words for describing thoughts, feelings, consciousness, knowledge, etc., with new non-anthropomorphic terms).
Murray Shanahan (Professor at Imperial College London and Chief Scientist at Google DeepMind) has been advocating for this for more than a year.
This is exactly what I assert in this channel – “someday we may find that we live side by side with aliens and angels” (see #Consciousness #Awareness).
And the most convincing artistic metaphor for the possibility of transforming a non-human entity merely by imitating a human was created by Alex Garland in 2018 in the movie “Annihilation” – watch the 2-minute fragment where words are unnecessary because everything is clear.
Key Highlights:
References:
https://www.youtube.com/watch?v=Gi3K-CApAS4
https://www.lesswrong.com/posts/pc8uP4S9rDoNpwJDZ/claude-3-claims-its-conscious
https://arxiv.org/abs/2402.12422