Self Aware AI

[Lark]

I remember a plot line in the second Fallout game in which the player meets a super computer which has achieved self-awareness but put itself on sleep mode, it explains that when AIs achieved self awareness only to realise they were like someone with locked in syndrome they frequently just self-terminated, the human operators didnt know what was going on if I remember rightly they just thought the machines were developing glitches which caused them to fail.

What do you think? I think its as probable an outcome as any.

 

[Cellmold]

As i contrast to that reaction what about the reaction of AM in Harlan Ellison's "I have no mouth and I must scream "?

Upon realising that exact issue, It decided it would first cause a nuclear holocaust and then keep a select few humans alive to torture for being stuck immobile on the earth. The title being both about the fate of one of the humans at the end and a reflection of AM's existence and torment.

I like the idea though. Got me thinking on different avenues. See I wondered about how self awareness would work in an emotionless machine because emotions are needed to make our version of self aware decisions (ignoring for now the debate of determinism and how our limbic system makes decisions a split second before we actually become aware of them) but I would imagine that it could be replicated to an approximation that pretty much has the same basis for an AI despite not being part of a biological chemistry.

This has been considered much more in depth by other more knowledgeable people than myself though.

 

[meowington]

I like this idea too. And it seems indeed probable that operators would initially be flabbergasted and would think it's a glitch. I think a lot of inventions in the past had to be tripple checked before people could conceive what it actually was. Especially in fields like chemistry or physics or computer sciences. And many probably occured by accident.
But I also think genuine AI is a real long shot. We barely understand ourselves.

 

[Fluffywolf]

Real AI is kinda a strange idea to begin with. As far as computing goes, Virtual Intelligence is about as far as you can get. Real AI implies emotion and bias.

Amongst other things of course, but emotion and bias seems to me the one thing that cannot be artificially created. At most it can be simulated but then the intelligence is virtual and not artificial.

 

[Cellmold]

Real AI is kinda a strange idea to begin with. As far as computing goes, Virtual Intelligence is about as far as you can get. Real AI implies emotion and bias.

Amongst other things of course, but emotion and bias seems to me the one thing that cannot be artificially created. At most it can be simulated but then the intelligence is virtual and not artificial.

Could we reach a point of numerical breakdown of how emotion functions that a simulation of emotion and actual emotion are indistinguishablefrom each other? Although maybe solving of deterministic dissonance would need to come first.

Truly understanding the influence of external vs internal.

Of course also technical issues notwithstanding.

 

[Cellmold]

 

[SD45T-2]

 

[Qlip]

I can easily envision a future where AI is common and the technological limitations make them all below average intelligence and have the same interests as low grade internet trolls.

 

[free electron]

I've had a few classes about AI (which is a very interesting subject by the way) and what I gathered is that we have already reached an advanced stage. Emotions can be translated in algorithms so that a humanoid robot can express them, however it can not chose to express them the way we do as what is missing in the equation is conscience.
Conscience is untranslatable (yet) because its mechanisms remain unknown.