G0W51

Im­mor­tal­ity: A Prac­ti­cal Guide

• G0W51 3 Mar 2015 1:23 UTC

  23 points

  on: Open thread, Mar. 2 - Mar. 8, 2015

  Perhaps it would be beneficial to use a unary numeral system when discussing topics on which biases like scope insensitivity, probability neglect, and placing too much weight on outcomes that are likely to occur. Using a unary numeral system could prevent these biases by presenting a more visual representation of the numbers, which might give readers more intuition on them and thus be less biased about them. Here’s an example: “One study found that people are willing to pay $80 to save || 1000 (2,000) birds, but only $88 to save |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| 1000 (200,000 birds).”

  Edit: Made it a bit easier to read.

• G0W51 26 Jan 2015 23:11 UTC

  13 points

  in reply to: Lumifer’s comment on: Im­mor­tal­ity: A Prac­ti­cal Guide

  The section on nutrition essentially is a “mainstream-advice data dump,” because I am not a professional nutritionist, so my judgments would likely be less accurate than mainstream advice. Analyzing each review or meta-analysis on diet individually would take far too long, and I didn’t want to simply link to the Harvard School of Public Health, because they give too much irrelevant (though interesting) information, and their articles seem somewhat poorly organized.

  Could you explain what’s wrong with place-holders? When I do happen to find articles explain topics accurately, I see no reason to repeat what they have already said.

  As for the eye-rolling parts, I kept them just in case somehow someone missed the obvious, which can happen. For those who didn’t miss the obvious, they can easily skim those parts.

• G0W51 5 Oct 2015 6:54 UTC

  8 points

  on: Open thread, Oct. 5 - Oct. 11, 2015

  What literature is available on who will be given moral consideration in a superintelligence’s coherent extrapolated volition (CEV) and how much weight each agent will be given?

  Nick Bostrom’s Superintelligence mentions that it is an open problem as to whether AIs, non-human animals, currently deceased people, etc should be given moral consideration, and whether the values of those who aid in creating the superintelligence should be given more weight than that of others. However, Bostrom does not actually answer these questions, other than slightly advocating everyone being given equal weight in the CEV. The abstracts of other papers on CEV don’t mention this topic, so I am doubtful on the usefulness of reading their entireties.

  (This is a repost.)

• G0W51 13 Apr 2015 23:10 UTC

  7 points

  on: Open Thread, Apr. 13 - Apr. 19, 2015

  Perhaps multiple choice tests in schools make people extra susceptible to privileging hypotheses. As a simplified example, if a student’s probability distribution to the answer of a question on a multiple choice test before seeing the choices is uniformly distributed amongst all integers from 1 to n, simply seeing an arbitrary integer as one of the, say, four options in a multiple choice test justifiably increases its probability of being correct to 0.25, a tremendous increase when n is large. Thus, on multiple choice tests, privileging a possible answer being “suggested” by being one of the answers the student can pick is a useful strategy on multiple choice tests, so we need to remember that privileging the hypothesis is not a viable strategy in other contexts. Maybe multiple choice tests aren’t so good after all...

• G0W51 19 Feb 2015 15:01 UTC

  7 points

  on: Open thread, Feb. 16 - Feb. 22, 2015

  Perhaps individuals are more liable to believe claims that have a small number of strong arguments than believe claims with a large number of weak arguments, even if the sum of the evidence for both claims is equally strong, as individuals often can’t hear all arguments for something, so they only hear the n strongest ones.

• G0W51 6 Apr 2015 21:35 UTC

  6 points

  on: Open Thread, Apr. 06 - Apr. 12, 2015

  I think I’ve found a somewhat easy-to-make error that could pose a significant existential risk when making an AGI. This error can potentially be found in hierarchical planning agents, where each high-level action (HLA) is essentially its own intelligent agent that determines what lower-level actions to do. Each higher-level action agent would treat determining what lower-level action to do as a planning problem and would try to take the action that maximizes its own utility function (if its a utility-based agent) or (if it’s a goal-based agent) probability of accomplishing its goal while minimizing its cost function (UOCF).

  For these agents, it is absolutely vital that each HLA’s UOCF prevents the HLA from doing anything to interfere with the highest-level action maximizing its utility function, for example by rewriting the utility function of higher-level actions or sending the higher-level actions deliberately false information. Failing to do so would result in an error that could significantly increase existential risk. To explain why, consider an agent whose highest-level action wants to maximize the number of fulfilling lives lived. In order to do this, the agent has a lower-level action whose goal is to go to a warehouse to get supplies. The cost function of this lower-level action is simply a function of, say, the amount of time it takes for the agent to reach the warehouse and the amount of money spent or money in damages done. In this situation, the lower-level action agent might realize that there is a chance that the higher-level action agent changes its mind and decides to do something other than go to the warehouse. This would cause the lower-level action to fail to accomplish its goal. To prevent this, this lower-level action may try to modify the utility function of the higher-level action to make it certain to continue trying to go to the warehouse. If this is done repeatedly by different lower-level actions, the resultant utility function could be quite different from the highest-level action’s original utility function and may pose a large existential risk. Even if the lower-level action can’t rewrite the utility function of higher-level actions, it may still sabotage the higher-level action in some other way to further its own goals, for example by sending false information to higher-level actions.

  To prevent this, the utility function of the lower-level action can simply be to maximize the highest-level action’s utility, and it can see the UOCF it was provided with as a rough method of maximizing the highest-level action’s utility function. In order to make the UOCF accurately represent the highest-level action’s utility function, it would (obviously) need to place high cost on interfering with the highest-level action’s attempts to maximize its utility. Some basic ideas on how to do this is for there to be very high cost in changing the utility functions of higher-level actions or giving them deliberately false information. Additionally, the cost of this would need to increase when the agent is more powerful, as the more powerful the agent is, the greater damage a changed utility function could do. Note that although higher-level actions could learn through experience what the UOCFs of lower-level actions should be, great care would need to be taken to prevent the AGI from, when still inexperienced, accidentally creating a lower-level action that tries to sabotage higher-level actions.

• G0W51 26 Jan 2015 23:43 UTC

  6 points

  in reply to: Fluttershy’s comment on: Im­mor­tal­ity: A Prac­ti­cal Guide

  I have a couple qualifications for your list. If one is vitamin D deprived (which is quite common), wearing sunscreen may do more harm than good. Also, as discussed in the original post, moderate alcohol consumption seems beneficial for some, so restricting alcohol consumption will only likely be beneficial if one drinks too much of it.

• G0W51 23 Sep 2015 4:23 UTC

  5 points

  on: Open thread, Sep. 21 - Sep. 27, 2015

  Where can one find information on the underlying causes of phenomena? I have noticed that most educational resources discuss superficial occurrences and trends but not their underlying causes. For example, this Wikipedia article discusses the happenings in the Somali Civil War but hardly discusses the underlying motivations of each side and why the war turned out how it did. Of course, such discussions are often opinionated and have no clear-cut answers, perhaps making Wikipedia a sub-optimal place for them.

  I know LW might not be the best place to ask this, but my intuition suggests that LWers may care more about this deeper-level understanding, so may be able to suggest resources.

• G0W51 7 Aug 2015 9:09 UTC

  5 points

  on: Open thread, Aug. 03 - Aug. 09, 2015

  I have heard (from the book Global Catastrophic Risks) that life extension could increase existential risk by giving oppressive regimes increased stability by decreasing how frequently they would need to select successors. However, I think it may also decrease existential risk by giving people a greater incentive to care about the far future (because they could be in it). What are your thoughts on the net effect of life extension?

• G0W51 20 Feb 2016 19:25 UTC

  4 points

  on: Open Thread Feb 16 - Feb 23, 2016

  Is there a term for a generalization of existential risk that includes the extinction of alien intelligences or the drastic decrease of their potential? Existential risk, that is, the extinction of Earth-originating intelligent life or the drastic decrease of its potential, does not sound nearly as harmful if there are alien civilizations that become sufficiently advanced in place of Earth-originating life. However, an existential risk sounds far more harmful if it compromises all intelligent life in the universe, or if there is no other intelligent life in the universe to begin with. Perhaps this would make physics experiments more concerning than other existential risks, as even if their chance of causing the extincion of Earth-originating life is much smaller than other existential risks, their chance of eliminating all life in the universe may be higher.

• G0W51 7 Aug 2015 17:33 UTC

  4 points

  in reply to: pcm’s comment on: Open thread, Aug. 03 - Aug. 09, 2015

  Oppression could cause an existential catastrophe if the oppressive regime is never ended.

• G0W51 6 Aug 2015 18:18 UTC

  4 points

  on: Open thread, Aug. 03 - Aug. 09, 2015

  The book Global Catastrophic Risks states that it does not appear plausible that molecular manufacturing will not come into existence before 2040 or 2050. I am not at all an expert on molecular manufacturing, but this seems hard to believe, given how little work seems to be going into it. I couldn’t find any sources discussing when molecular manufacturing will come into existence. Thoughts?

• G0W51 20 Jun 2015 7:17 UTC

  4 points

  on: Open Thread, Jun. 15 - Jun. 21, 2015

  What are some recommended readings for those who want to decrease existential risk? I know Nick Bostrom’s book Superintelligence, How can I reduce existential risk from AI?, and MIRI’s article Reducing Long-Term Catastrophic Risks from Artificial Intelligence are useful, but what else? What about non-AI-related existential risks?

• G0W51 28 Jan 2015 0:20 UTC

  4 points

  in reply to: Fluttershy’s comment on: Im­mor­tal­ity: A Prac­ti­cal Guide

  Could you give sources for your claims?

  As for the one source you did give, which was Wikipedia, it may have been referring to the average effect on expectancy, which, given the sizable frequencies of alcoholism and drunk driving, very well could be negative. Wikipedia gave two citations for the claim, but one of them wasn’t available. The other one was a brief mention on a powerpoint slide from Harvard that gave no further explanation. Ironically, I got my information on alcohol’s effect on death from Harvard as well.

• G0W51 25 Jan 2015 16:02 UTC

  4 points

  on: Open thread, Oct. 6 - Oct. 12, 2014

  I’m having trouble posting an article, so any help would be appreciated. I’ve tried to make a discussion post, and whenever I click “submit,” it says, “submitting...” for a few seconds. However, the thread never appears in the discussion section. Also, when I try to close out of the submit article page, it says, “you’ve made changes to the article, but haven’t submitted it.” Does anyone have any idea what could be causing this? My only idea is that the target and class of hyperlinks aren’t set, or that the article’s too long; it’s 27 pages.

My Skepticism

• G0W51 31 Jan 2015 18:33 UTC

  3 points

  on: Open thread, Jan. 26 - Feb. 1, 2015

  It’s worth estimating when existential risks are most likely to occur, as knowing this will influence planning. E.g. If existential risks are more likely to occur in the far future, it would probably be best to try to invest in capital now and donate later, but if they are more likely to occur in the near future, it would probably be best to donate now.

  So, what’s everyone’s best estimates on when existential catastrophes are most likely to occur?

• G0W51 29 Jan 2015 0:19 UTC

  3 points

  in reply to: Fluttershy’s comment on: Im­mor­tal­ity: A Prac­ti­cal Guide

  I did more research, and the results I found support my statements on cellphones and alcohol.

  Could you rephrase your last paragraph? It’s missing a word. :)

• G0W51 28 Jan 2015 0:56 UTC

  3 points

  in reply to: Salemicus’s comment on: Im­mor­tal­ity: A Prac­ti­cal Guide

  “‘Assuming individuals receive some finite, above zero amount of utility per unit time, living forever would give infinitely >more utility than living for a finite amount of time.’

  Only if the utility per unit time is constant. ” You’re right. Edited.

  Most useful things are subject to diminishing returns. Why should lifespan be an exception? I could even easily imagine that at a certain point, utility turns negative.

  Do you ever get bored, or find yourself at a loose end? Do you think that would happen more or less often if your >lifespan were longer? I think it would happen much less, if at all, if lifespan was longer, as in the future, one could potentially cure boredom, as preventing boredom sounds much easier than living forever.

  Human lifespan (and health in old age) has increased considerably over the last 100 years in the Western world, and >that increased lifespan has largely been spent in retirement. Consider how retired people typically spend their time, >and ask yourself whether that lifestyle strikes you as appealing. Taking the outside view, why do you think you would >use your increased lifespan in any more productive a manner? As far as I know, the elderly are unproductive because agings has degraded their physical and mental abilities so far that they are no longer very useful in the work force. This wouldn’t happen if aging was cured. Regardless, one wouldn’t need to be productive one AGI’s created.

  I highly suggest reading the fun theory sequence.