red75

• red75Sep 7, 2010, 8:43 PM

  3 points

  on: Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread, part 3

  Godric hadn’t told anyone, nor had Rowena if she’d known; there might have been any number of wizards who’d figured it out and kept their mouths shut.

  I’m still can’t figure out what’s dangerous about sharing that knowledge. My obviously unsatisfying guesses in no particular order:

  *Making yet unreachable sour grapes of immortality more sweet.

  *Harry’s concern for self-awareness of patronus v2.0.

  *Outlaws will be better protected from law enforcement.

  *Increased chances of Azkaban break out.

  *Disclosure of Harry’s unique power.

  *Extinction of dementors.

• red75Aug 24, 2010, 8:52 PM

  0 points

  in reply to: Perplexed’s comment on: Min­i­mum com­pu­ta­tion and data re­quire­ments for con­scious­ness.

  Self-model theory of subjectivity can also suggest

  (7) Ability to explicitly represent state of our own knowledge, intentions, focus of attention etc. Ability to analyse performance of our own brain and find ways to circumvent limitations. Ability to control brain’s resource allocation by learned (vs evolved) procedures.

  The consciousness stuff is about thinking that maps really match up to territory.

  Interesting thing about consciousness is that the map is a part of territory it describes, and as the map should be represented by neuronal connections and activity it can presumable influence territory.

• red75Aug 16, 2010, 8:07 AM

  4 points

  in reply to: Perplexed’s comment on: Ex­is­ten­tial Risk and Public Relations

  4 . And hence an effective safeguard would be to simply not give the machine its own credit card!

  (Powerful) optimization processes can find such ways of solving problems by exploiting every possible shortcut that it is hard to predict those ways in advance. Recently here was an example of that. Genetic algorithm found unexpected solution of a problem exploiting analog properties of particular FPGA chip.

• red75Aug 13, 2010, 6:23 AM

  0 points

  in reply to: Eliezer Yudkowsky’s comment on: Should I be­lieve what the SIAI claims?

  There is something that makes me feel confused about MWI. Maybe it is its reliance on anthropic principle (probability of finding myself in a world where recorded history have probability P (according to Born’s rule) must be equal to P). This condition depends on every existing universe, not just on ours. Thus it seems that to justify Born’s rule we should leave observable evidence behind and trail along after unprovable philosophical ideas.

• red75Aug 3, 2010, 8:17 PM

  1 point

  in reply to: Blueberry’s comment on: Ra­tion­al­ity quotes: Au­gust 2010

  Thanks. It is identical argument modulo my inability to make all reasoning and premises sufficiently transparent.

• red75Aug 3, 2010, 6:11 PM

  1 point

  in reply to: thomblake’s comment on: Ra­tion­al­ity quotes: Au­gust 2010

  I am not establishing qualia as physical properties of brain’s activity, I think of them as descriptions of specific neural activity in the terms of human’s self-model. And limitations of that self-model (it’s not sufficiently detailed to refer to individual neurons) don’t allow to establish unambiguous correspondence between physical description of brain and self-model description of brain within that self-model.

  Mary knows that too and can perform an experiment to activate those neurons without having a ‘red thing’ in front of her, using her incredible superhuman intelligence and resources.

  And what is a difference between seeing red thing and activation of those neurons? The point of “Mary’s room” is to know what seeing red means without actually seeing it.

• red75Aug 3, 2010, 5:39 PM

  3 points

  in reply to: orthonormal’s comment on: Ra­tion­al­ity quotes: Au­gust 2010

  Intuitions don’t matter. If Mary can’t activate her neural pathways participating in creation of experience of seeing red, then she has no means of knowing how she will experience redness. All models she can create in her mind will be external to her as the mind created by actions of human being in Chinese room is external to that human being.

  And this limitation of our conscious understanding feels to us like a thing we have which cannot in principle be reduced.

  It is not only conscious understanding that is required, we will need a conscious control of individual neurons and synapses to be able to experience qualia given just a description of it. For example, to be able to name color and imagine color given its name, Mary (roughly speaking) should manually connect neurons in her visual cortex to the neurons in her Broca’s area and to the neurons in her auditory cortex.

  So I think that, contrary to Dennet, Mary will get new information when she will see colors, as human’s brain construction doesn’t allow to acquire that information by other means. Thus in a sense human’s qualia cannot be reduced.

• red75Aug 2, 2010, 2:10 PM

  0 points

  in reply to: humpolec’s comment on: Open Thread, Au­gust 2010

  Yes, something like that. But I don’t think that consciousness of such being will be dramatically different, because it still should contain “central executive” that still coordinates overall behavior of that being and still controls direction and distribution of attention that is however much more fine-grained than human’s one.

• red75Aug 2, 2010, 1:03 PM

  0 points

  in reply to: KrisC’s comment on: Open Thread, Au­gust 2010

  I rather say higher level functions is excessively redundant. Then there are coordination problems, competition for shared resources (e.g. money, sexual partner), possibly divergence of near- and far-term goals, relatively low in-group communication speed, possibly less number of cross-domain-of-knowledge insights.

• red75Aug 1, 2010, 7:56 PM

  0 points

  in reply to: humpolec’s comment on: Open Thread, Au­gust 2010

  Waste of processing power. Having dozens of focuses of attention and corresponding body/​brain construction is more efficient.

• red75Aug 1, 2010, 12:09 PM

  2 points

  in reply to: AlephNeil’s comment on: AI co­op­er­a­tion in practice

  On the other hand the fact that proves(“A()==”Cooperate” && B()==”Cooperate”″) was called means that A()==”Defect” && B()==”Cooperate” is unprovable under maxDC proof length. The question is can this fact be used in proof. The proof checker can be programmed to add such self-evident propositions into set of axioms, but I can’t clearly see consequences of this. And I have a bad feeling of mixing meta- and object levels.

• red75Aug 1, 2010, 10:29 AM

  2 points

  in reply to: AlephNeil’s comment on: AI co­op­er­a­tion in practice

  Can this problem be solved by adjusting maximal length of checked proofs such that unprovability of proves(“A()==”Defect” && B()=”Cooperate”″, maxDC) is provable by proves(..., maxsteps)?

• red75Jul 31, 2010, 6:04 PM

  2 points

  in reply to: Vladimir_Nesov’s comment on: AI co­op­er­a­tion in practice

  On the other hand, pathological formal language can be constructed, that makes proof of “main()==1” arbitrary large (e.g. where 1 is represented by 3^^^3 symbols). Thus it is better to be proven that formal language isn’t pathological, i.e there is no prover proves’ that satisfies condition

  \exists n1 \exists n2 \forall code (n1<n2 & proves'(code,n1)=proves(code,n2) )
  

• red75Jul 31, 2010, 4:38 PM

  2 points

  in reply to: Vladimir_Nesov’s comment on: AI co­op­er­a­tion in practice

  Ouch. maxsteps means maxprooflenght. My inattention, sorry.

• red75Jul 31, 2010, 4:31 PM

  −2 points

  in reply to: Vladimir_Nesov’s comment on: AI co­op­er­a­tion in practice

  Enumeration order of proofs is not specified, so we can construct pathological prover, that reorders proofs such that all proofs of “main()==1” require more steps than maxsteps.

• red75Jul 28, 2010, 7:20 PM

  0 points

  in reply to: Vladimir_Nesov’s comment on: Me­taphilo­soph­i­cal Mysteries

  Does “extremal counterfactual mugging” where $100 is replaced by self-destruction of agent and $10000 is replaced by creation of 100 agent’s copies (outside of agent’s light cone) requires same answer as counterfactual mugging?

• red75Jul 28, 2010, 6:11 PM

  2 points

  on: Alien par­a­site tech­ni­cal guy

  (The magic was the tricky part.)

  Are you sure that making that part you didn’t implement prosthetic consciousness? Suppression of undesirable involuntary intentions is a part of consciousness’ job as it seems.

• red75Jul 28, 2010, 5:20 PM

  0 points

  in reply to: NancyLebovitz’s comment on: Harry Pot­ter and the Meth­ods of Ra­tion­al­ity dis­cus­sion thread

  BTW. Eliezer removed these words. Now they are ” As soon as I’m old enough”.

• red75Jul 26, 2010, 4:35 PM

  1 point

  in reply to: Vladimir_Nesov’s comment on: Con­trived in­finite-tor­ture sce­nar­ios: July 2010

  I am still puzzled how preference corresponds to the physical state of brain. Is preference only partially presented in our universe (intersection of set of universes which correspond to your subjective experience and set of universes which correspond to mine subjective experience)?

• red75 23 Jul 2010 17:29 UTC

  3 points

  in reply to: Nisan’s comment on: Sim­plified Hu­man­ism, Pos­i­tive Fu­tur­ism & How to Prevent the Uni­verse From Be­ing Turned Into Paper Clips

  P(extinction-event)~=P(realized-other-extinction-threat)+P(hand-coded-CEV/​FAI-goes-terribly-wrong)+P(AGI-goes-FOOM)

  P(AGI-goes-FOOM)~= 1 - \prod j [P(development-team-j-will-not-create-AGI-before-FAI-is-developed) + {1-P(development-team-j-will-not-create-AGI-before-FAI-is-developed) } P(development-team-j-can-stop-AGI-before-FOOM) ]

  So strategy is to convince every development team, that no matter what precautions they use P(development-team-j-can-stop-AGI-before-FOOM)~=0. And development of recommendations for AGI containment will suggest that P(development-team-j-can-stop-AGI-before-FOOM) can be made sufficiently high, thus lowering P(development-team-j-will-not-create-AGI-before-FAI-is-developed). Given overconfidence bias it is plausible to assume that latter will increase P(AGI-goes-FOOM).

  I withdraw suggestion.