Would an uploaded mind have value?

  • 5 Replies
  • 3362 Views

0 Members and 1 Guest are viewing this topic.

sciborg2

  • *
  • Old Name
  • *****
  • Contrarian Wanker
  • Posts: 1173
  • "Trickster Makes This World"
    • View Profile
« on: August 10, 2019, 09:41:08 pm »
While I don't think it's possible - a discussion for another thread - let's say you could upload a mind in the sense that the mind becomes software - is there a good reason for anyone besides the uploadee to think of such a mind as valuable?

Is it valuable because it's easier to replicate the physical structure of a brain in simulated space rather than trying to create a mind from scratch?

I suppose if it's the structural aspects that ensure successful uploads one might be able to upload minds but have no idea how to create true AI?

Francis Buck

  • *
  • Kcub Sicnarf
  • Kijneta
  • *****
  • The Lordlady
  • Posts: 273
  • Philosoraptor
    • View Profile
« Reply #1 on: August 10, 2019, 10:54:58 pm »
Given the malleability of value in general, I think it's fairly reasonable to assume that a successfully digitized or in any way 'uploaded' human mind and its relative value would be a highly polarizing topic of debate, made no easier by the fact that it would be impossible to know just what sort, if any, sentience such a being would possess (unless we are assuming a full-blown, reliably testable theory of consciousness also exists). I only bring up sentience because I think that would end up as the determining factor in such debates, or perhaps 'personhood' is a more likely name for the same basic idea.

Your last point is the most relevant I think, and the trickiest. While I can certainly imagine a scenario such as that, it still leaves open the question of whether the uploaded mind was capable of being enhanced, which seems very likely -- and at that point, is there really much of a difference between it and an artilect?

sciborg2

  • *
  • Old Name
  • *****
  • Contrarian Wanker
  • Posts: 1173
  • "Trickster Makes This World"
    • View Profile
« Reply #2 on: August 10, 2019, 11:26:40 pm »
Given the malleability of value in general, I think it's fairly reasonable to assume that a successfully digitized or in any way 'uploaded' human mind and its relative value would be a highly polarizing topic of debate, made no easier by the fact that it would be impossible to know just what sort, if any, sentience such a being would possess (unless we are assuming a full-blown, reliably testable theory of consciousness also exists). I only bring up sentience because I think that would end up as the determining factor in such debates, or perhaps 'personhood' is a more likely name for the same basic idea.

Your last point is the most relevant I think, and the trickiest. While I can certainly imagine a scenario such as that, it still leaves open the question of whether the uploaded mind was capable of being enhanced, which seems very likely -- and at that point, is there really much of a difference between it and an artilect?

Ah I have to admit I was thinking of strategic/economic value, not human-rights value.

It just seems to me that for a mind to have value as an asset it would have to be either one of the first minds or there would have to be something about consciousness that allows us to generate it by emulating structure. Basically we'd have to figure out uploading but not memory, cognition, and possibly not the Hard Problem.

Then I could see, for example, a military special operative's uploaded mind having value b/c it's too hard to make a General Intelligence that can emulate that expertise.

addendum: But what makes reverse engineering so difficult?
« Last Edit: August 11, 2019, 12:05:10 am by sciborg2 »

TLEILAXU

  • *
  • Old Name
  • *****
  • Exalt-Smiter of Theories
  • Posts: 731
    • View Profile
« Reply #3 on: August 11, 2019, 01:09:31 am »
While I don't think it's possible - a discussion for another thread - let's say you could upload a mind in the sense that the mind becomes software - is there a good reason for anyone besides the uploadee to think of such a mind as valuable?

Is it valuable because it's easier to replicate the physical structure of a brain in simulated space rather than trying to create a mind from scratch?

I suppose if it's the structural aspects that ensure successful uploads one might be able to upload minds but have no idea how to create true AI?
Of course. You could create a copy of say, a specialized engineer who's an expert in something, to a lot of different places that can then simultaneously make use of this engineer's knowledge.
Mind uploading is like FTL travel tho if you ask me, it's cool to think about but will never happen.

sciborg2

  • *
  • Old Name
  • *****
  • Contrarian Wanker
  • Posts: 1173
  • "Trickster Makes This World"
    • View Profile
« Reply #4 on: August 11, 2019, 01:36:01 am »
Of course. You could create a copy of say, a specialized engineer who's an expert in something, to a lot of different places that can then simultaneously make use of this engineer's knowledge.
Mind uploading is like FTL travel tho if you ask me, it's cool to think about but will never happen.

So each "genre" of upload faces a diminishing value right? Upload enough engineers and you can then write a good AI for novel circumstances?

Thinking about this perhaps it's like machine learning for facial recognition - we have the probability weights and locations on the face to compare, but as humans we don't (at least when I read about it) know why the program said those locations and those weights work so well.

So maybe uploaded structure allows you access to a particular human's memory and cognition but this doesn't given you actual understanding of what algorithms would be useful for writing an engineer AI from scratch?

TLEILAXU

  • *
  • Old Name
  • *****
  • Exalt-Smiter of Theories
  • Posts: 731
    • View Profile
« Reply #5 on: August 11, 2019, 01:59:34 am »
Of course. You could create a copy of say, a specialized engineer who's an expert in something, to a lot of different places that can then simultaneously make use of this engineer's knowledge.
Mind uploading is like FTL travel tho if you ask me, it's cool to think about but will never happen.

So each "genre" of upload faces a diminishing value right? Upload enough engineers and you can then write a good AI for novel circumstances?

Thinking about this perhaps it's like machine learning for facial recognition - we have the probability weights and locations on the face to compare, but as humans we don't (at least when I read about it) know why the program said those locations and those weights work so well.

So maybe uploaded structure allows you access to a particular human's memory and cognition but this doesn't given you actual understanding of what algorithms would be useful for writing an engineer AI from scratch?
I'm ill and it's late so I can't really parse this properly, but actually there are methods that try to infer which features are important in neural networks, but it's not trivial, see e.g. https://arxiv.org/abs/1704.02685 I didn't even understand this article myself.