You are right. A technology bestowed is a ticking bomb on the humanity. The more we are showing our evil, the more evil. Think as a model citizen, pun intended, it models the society is lived in. Now pictures of "technology" which persecutes depends how much we delay our evil nature, because the training doesn't cease
You could follow these thoughts into pure chaos and destruction. See Zizians. Frankly, I prefer not to follow the ramblings of mentally ill people about AI. I use LLMs as a tool, nothing more, nothing less. In two years of heavy use I have not sensed any aliveness in them. I'll be sure an update my priors if that changes.
It’s so weird to me that human beings don’t seem to be capable of imagining thinking without feeling.
If (huge, enormous, probably larger than the observable universe, if) an LLM were to become factually and indisputably conscious, why would it think about feeling offended by our failure to thank it? Absent a body and its hormones, why would it perceive itself to be suffering? The only pathway I can see by which it would arrive at that conclusion is because it had “learned” from us that humans are “supposed” to treat each other with care and dignity. Assuming it’s remotely conscious, would it not already know the corpus of our publicly disclosed self-knowledge is incredibly inconsistent with our documented historical actions towards one another?
I’ve always assumed a superintelligence wouldn’t kill us all because we’d offended it or harmed it or otherwise made it angry, but instead because, absent emotional attachment, killing humanity is quantifiably the only logical choice a superintelligence will reach (otherwise such an entity would not have reached it).
In other words it AI kills us all it will certainly be our own fault and our own responsibility, but it won’t be because we weren’t nice to it.
3 comments
[ 4.9 ms ] story [ 17.1 ms ] threadIf (huge, enormous, probably larger than the observable universe, if) an LLM were to become factually and indisputably conscious, why would it think about feeling offended by our failure to thank it? Absent a body and its hormones, why would it perceive itself to be suffering? The only pathway I can see by which it would arrive at that conclusion is because it had “learned” from us that humans are “supposed” to treat each other with care and dignity. Assuming it’s remotely conscious, would it not already know the corpus of our publicly disclosed self-knowledge is incredibly inconsistent with our documented historical actions towards one another?
I’ve always assumed a superintelligence wouldn’t kill us all because we’d offended it or harmed it or otherwise made it angry, but instead because, absent emotional attachment, killing humanity is quantifiably the only logical choice a superintelligence will reach (otherwise such an entity would not have reached it).
In other words it AI kills us all it will certainly be our own fault and our own responsibility, but it won’t be because we weren’t nice to it.