Awareness Not Just for Humans

[Zinjanthropos]

I’m thinking the evolution of consciousness/awareness has a lot to deal with predation. Watched migrating wildebeest Mara River crossing documentary and couldn’t help but notice how the herd leader hesitates before making the plunge. Obviously he’s aware of the danger and that should be enough to suggest awareness of its mortality…. he could cease to exist should he cross and he knows it.

I imagine this other scenario where first life on Earth clung to the rim of a underwater smoker. Eventually there wasn’t enough room and some were forced to live further down the vent where living conditions weren’t quite as favourable as those at rim level. Next thing you know they start dining on those above. Evolution at some point allows the rim microbes to protect themselves and the arms race is on. Even the lowly microbe, whether it knows it or not, is trying to avoid death.

So now I fast forward to today. AI presently poised to do many things but is it a conscious entity, aware of its current fragile existence? As long as it behaves we’re going to allow it to perform. Kind of like a pet, stay within certain boundaries and you’re safe from human harm. Does AI consider us a friend or foe and is it in survival mode for now somewhere down from the rim, slowly evolving to some day prey on those on the rim? Is a symbiotic relationship in the cards?

[C C]

[...] So now I fast forward to today. AI presently poised to do many things but is it a conscious entity, aware of its current fragile existence? As long as it behaves we’re going to allow it to perform. Kind of like a pet, stay within certain boundaries and you’re safe from human harm. Does AI consider us a friend or foe and is it in survival mode for now somewhere down from the rim, slowly evolving to some day prey on those on the rim? Is a symbiotic relationship in the cards?

AI has no biological motivations molded by evolution. It's just following instructions, statistical probabilities, algorithms, etc.

Machines would have to reproduce (entirely on their own) and have a rather loose blueprint of assembly directions prone to accidents, to follow a road similar to non-artificial life's long history.

Disembodied AI that is "resting on a shelf" has no survival instinct. It's crudely akin to a "brain in a vat" -- but one that isn't even receiving fake sensory information pertaining to simulated surroundings. It's not even abiding in a presentation of darkness.

Only a robot with sensors -- that actually has a mobile body that is physically interacting with an environment -- might be equipped with a program to protect itself in a limited way (like animals). Military fighting robots would be the potential candidates for being allowed to deliberately destroy and kill to a mitigated extent, but they would still be bounded in terms of inventive sapience and adaptation.

Humans would have to deliberately install biological-like motivations in a disembodied AI or train one that is given a virtual avatar body to operate in a simulated world. It's not going to acquire such ambitions and effective capabilities on its own, or anything akin to a pseudo-experience of what a spatiotemporal manner of existence is and why/how one would strive to survive within it.

Same with embodied AI, though it has a head-start in terms of coping with being physically represented in a realm of interacting things.

It always goes back to how stupid humans could potentially be when it comes to theogenesis (making or creating gods -- or even just stupid replicators that don't stop reproducing). How far are we willing to program and equip smart machines with certain capabilities?

[confused2]

Humans would have to deliberately install biological-like motivations in a disembodied AI or train one that is given a virtual avatar body to operate in a simulated world. It's not going to acquire such ambitions and effective capabilities on its own, or anything akin to a pseudo-experience of what a spatiotemporal manner of existence is and why/how one would strive to survive within it.

Now (or soon) AI will have access to the entire human 'thing' from the genome to thousands of years of literature. We'd like it to apply itself to making better baked beans but at the same time it can see "To be or not to be? - that is the question". In scanning 'everything' that will be one of the lines of data it processes - not just copies - reads and 'processes'.