[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math


View post   

File: 13 KB, 657x527, 1547453157060.png [View same] [iqdb] [saucenao] [google]
10358130 No.10358130 [Reply] [Original]

Say that a general ai was made that met the requirements to be considered sentient. If you took the set of instructions that the computer program runs and followed the instructions all the same on paper (regardless of however long it takes) would it still be considered a sentient being? What would it be that was sentient?

>> No.10358135

You’re rephrasing the “Chinese Room”, right? It’s a baffling thing to me that people think this is profound. YOU are the sentient being performing the instructions.

>Say that a general ai was made that met the requirements to be considered sentient.

That wouldn’t be hard. We could probably do it now. Just make an ant brain.

>> No.10358136

We are never going to be able to verify if an AI (or any other system, for that matter) experiences sentience. Sentience is purely subjective and not falsifiable

>> No.10358146

>>10358130
the mapping from In to Out

>> No.10358149

>>10358136
Anything that is at least sort of homologous to animal neurology works as far as I’m concerned.

>> No.10358159

>>10358135
So you are saying a sentient being's set of instructions have to be ran by non sentient parts? Why does it matter either way how the instructions are being done if they both follow to same logic so that the things feeding in data were indistinguishable?

>> No.10358166

>>10358146
Shouldn't all of a computers processes be able to be replicated by hand if they all follow some string of logic though? Wouldn't it just be a matter of treating the set of input in the same sort of way?

>> No.10358205

>>10358159
There’s no “set of instructions”. Neural processes are active and ongoing interactions between neurons that constantly adjust themselves and engage in autoperception. Their behavior is organic rather than set like the line of code involved in pressing a button on a control panel and a door opening elsewhere. That’s what the reflexive, sensory, and motor parts of the nervous system are like. What you’d do to create a nonorganic entity that, as far as we know, also experiences qualia is to make a homologous system that can receive sensory information from, say a camera and microphone, and process this data, potentially storing it, in a plastic system of binary pseudoneurons. Robot Me could probably store a video recording of the entire day but, like the actual brain, would strip it down and compress it, potentially deleting it entirely if it’s not useful. This is the principle difference between a hypothetical calculator and a brain. The calculator isn’t doing a damn thing when not in use, and it’s coding is purely static. A brain is constantly firing and changing from the time the first neurons form in the embryo and keeps doing that until death. It can observe internally stored data and reflect on it, or use compiled information to synthesize new experiences like we do in dreams.

>> No.10358252

>>10358130
this assumes that a computer does things one instruction at a time which only holds for one type and one level of abstration. You can say of your own feeling self, if there's demon that stops time every nanosecond and goes and carries out one by one the interactions that would have got us to the next nanoseconds state naturally, then increments time 1 nanosecond, this is obviously imperceptible to you at your operating level of abstraction but now your same question of "are you really sentient?" applies exactly equivalently as it does to your 'ai'
Therefore if yes to demon/human system,
Yes to ai/von-neumann instructions system