The_Jaded_One

• The_Jaded_One 20 Nov 2014 21:15 UTC

  −3 points

  in reply to: examachine’s comment on: Musk on AGI Timeframes

  +1 for entertainment value.

  EDIT: I am not agreeing with examachine’s comment, I just think it’s hilariously bad.

• The_Jaded_One 20 Dec 2014 21:33 UTC

  0 points

  on: Su­per­in­tel­li­gence 14: Mo­ti­va­tion se­lec­tion methods

  When we jump from direct normativity to indirect normativity, it is reasonably claimed that we gain a lot.

  I sometimes wonder whether the issue of indirect normativity has been pushed far enough. The limiting case is that there is some way to specify, in “machine comprehensible” terms, that a software intelligence should “do what I want”.

  “outsourcing the hard intellectual work to the AI”

  Just how much can be outsourced?

  Could you program a software intelligence to go read books like Superintelligence, understand the concept of “Friendliness” or “Motivational alignment”, and then be friendly/​motivationally aligned with yourself?

  And couldn’t the problem of selecting a method to compromise between the billions of different axiologies of the humans on this planet be outsourced to the AI by telling it to motivationally align with the team of designers and their backers, subject to whatever compromises would have been made had the team tried to directly specify values? This is not to say I am advocating a post-singleton world run purely for the benefit of the design team/​project, but that if such a team (or individual) were already committed to trying to design something like “The CEV of humanity”, then an AI which was motivationally aligned with them would continue that task more quickly and safely.

  Anyway, I think there is a fruitful discussion to be had thinking about how the maximum amount of work can be offloaded to the AI; perhaps work on friendly AI should be though of as that part of the motivational alignment problem that simply has to be done by the human(s).

• The_Jaded_One 4 Jan 2015 15:04 UTC

  4 points

  on: Ten­ta­tive Thoughts on the Cost Effec­tive­ness of the SENS Foundation

  “about ten times the number of people who die each year would live 30 years of healthy life that they wouldn’t have lived otherwise.”

  This is wrong—those people get to live for thousands of years of SENS works. Or possibly even more.

• The_Jaded_One 4 Jan 2015 15:25 UTC

  6 points

  on: Ten­ta­tive Thoughts on the Cost Effec­tive­ness of the SENS Foundation

  $0.059 versus $2790 is a massive difference. I think what this mostly reflects is that SENS is claiming to make a massive difference in health outcomes, whereas contemporary biomedical research is much more modest in its ambition.

  The real question is what chance SENS has of succeeding. I’m unfortunately not enough of a domain expert to have a good idea about this, but over the years I have become increasingly skeptical; it seems to me that SENS is a great idea that will fail for a bunch of interesting reasons.

• The_Jaded_One 4 Jan 2015 17:47 UTC

  0 points

  on: Su­per­in­tel­li­gence 16: Tool AIs

  The fundamental problems with “tool AI” are

  (1) The difficulty of specifying what it means for a superintelligent system to only have a limited or bounded effect.

  (2) The difficulty of getting an oracle AI to “just tell the truth” when the questions you ask it don’t have a simple, straight answer that can be told, and where the choices about what to say will have significant effects upon the world such that you would be better off explicitly having a sovereign AI because your oracle is now moonlighting as a sovereign when it chooses how to answer.

• The_Jaded_One 4 Jan 2015 23:28 UTC

  4 points

  in reply to: Fluttershy’s comment on: Ten­ta­tive Thoughts on the Cost Effec­tive­ness of the SENS Foundation

  I have specific ideas about why it will fail: the various interventions to combat the 7 damage types will all meet with complications and unexpected difficulties.

  WILT, for example, might kill people or animals for years and years whilst the bugs are ironed out because they won’t get the cell replacement therapy right.

  MitoSENS will unearth the complexities of mitochondrial DNA and why it’s hard to move it into the nucleus.

  etc, etc.

  But as we do this work, we will learn a lot about the body and there will be some amazing spinoffs. For example LysoSENS and AmyloSENS will probably help with age related diseases. Eventually it might even come together to extend lifespan for another 50 years until the next unforseen problem crops up. But I expect that to take at least till 2090, and a lot will have changed by then. By 2090, cryo, AI and uploading will probably look a lot better than being the next human lab-rat for SENS.

• The_Jaded_One 4 Jan 2015 23:33 UTC

  3 points

  in reply to: Fluttershy’s comment on: Ten­ta­tive Thoughts on the Cost Effec­tive­ness of the SENS Foundation

  On the subject of SENS vs givewell, I think givewell have already kind of abandoned the goal of rational philanthropy in the way that they dismissed x-risk and continue to pour money into developing world aid. SENS could at least change the world eventually and affect billions/​trillions of lives, whereas givewell’s third world aid is always going to be a tiny drop in a gargantuan ocean.

• The_Jaded_One 5 Jan 2015 11:39 UTC

  −1 points

  in reply to: bogus’s comment on: Ten­ta­tive Thoughts on the Cost Effec­tive­ness of the SENS Foundation

  since it’s acting to lift these societies from a bad equilibrium

  If I agreed with this then I’d be more positive about givewell. I think it’s wishful thinking, though. In reality, societies in Africa are not in a bad equilibrium primarily because of malaria or malnutrition, they’re in a bad equilibrium because of the backwards values that African people hold, such as loyalty to the tribe/​extended family rather than the state, lack of support for western values like accountable government, basic rationality, equality under the law, fair enforcement of contracts, etc. We don’t hear much about this because it doesn’t fit with the political narrative of the kind of people who spend their time trying to help the third world.

  Malnutrition is the visible surface symptom of “these are uncivilised, backwards people caught in a series of petty tribal wars”.

  EDIT: Just let me disclaim that this is not supposed to be an excuse to not help the developing world. I think we should help them, but not by giving out food or bed nets or medicine. They need better political and value systems, which we could (and should) give them via charter cities.

• The_Jaded_One 5 Jan 2015 15:19 UTC

  2 points

  in reply to: bogus’s comment on: Ten­ta­tive Thoughts on the Cost Effec­tive­ness of the SENS Foundation

  So I looked up the poorest countries in the world. Depending on how you measure it, Malawi, the DR of congo and the Central African Republic are at the bottom.

  Wikipedia on the Democratic Republic of Congo:

  DR Congo is extremely rich in natural resources but political instability, a lack of infrastructure and a culture of corruption have limited development, extraction and exploitation efforts

  The Central African Republic

  The Central African Republic Bush War began in 2004 and, despite a peace treaty in 2007 and another in 2011, fighting broke out between government, Muslim, and Christian factions in December 2012, leading to ethnic and religious cleansing

  Malawi:

  Corruption within the government is seen as a major issue, despite the Malawi Anti-Corruption Bureau’s (ACB) attempts to reduce it. The ACB appears to be successful at finding and prosecuting low level corruption, but higher level officials appear to be able to act with impunity

• The_Jaded_One 8 Jan 2015 16:15 UTC

  0 points

  in reply to: CellBioGuy’s comment on: Ten­ta­tive Thoughts on the Cost Effec­tive­ness of the SENS Foundation

  “because poking a complicated system will often just make it worse.”

  yeah that was my view. I have only an amateur interest in biology though.

• The_Jaded_One 8 Jan 2015 17:51 UTC

  1 point

  in reply to: Mark_Friedenbach’s comment on: Ten­ta­tive Thoughts on the Cost Effec­tive­ness of the SENS Foundation

  It can’t be a few hundred because that would imply that the non-natural-causes death rate is about 20-30% of the natural-causes death rate, which isn’t true. At least in the developed world.

  I think it’s more like a few thousand.

  And driverless cars are on the way, which would reduce the death rate drastically.

• The_Jaded_One 20 Jan 2015 8:34 UTC

  −1 points

  in reply to: Lumifer’s comment on: Ten­ta­tive Thoughts on the Cost Effec­tive­ness of the SENS Foundation

  if P(life length=x) = p(1-p)^(x-1) with p=0.001, then E(life length) = 1/​p = 1000. It’s a geometric random variable.

  Which is NOT A FEW HUNDRED YEARS

(mis­lead­ing ti­tle re­moved)

• The_Jaded_One 29 Jan 2015 7:25 UTC

  1 point

  in reply to: jsteinhardt’s comment on: (mis­lead­ing ti­tle re­moved)

  I disagree that it is as inaccurate as you claim. Specifically, they did actually say that “AI doomsday scenarios belong more in the realm of science fiction”. I don’t think it’s inaccurate to quote what someone actually said.

  When they talk about “having more work to do” etc, it seems that they are emphasizing risks of sub-human intelligence and de-emphasizing the risks of superintelligence.

  Of course LW being LW I know that balance and fairness is valued very highly, so would you kindly suggest what you think the title should be and I will change it.

  I will also add in the paragraphs you suggest.

• The_Jaded_One 29 Jan 2015 7:29 UTC

  6 points

  in reply to: Vaniver’s comment on: (mis­lead­ing ti­tle re­moved)

  AI research field has this under control—we recognize that safety is super important and are working hard on it

  Great, except

  a) they don’t have it under control and

  b) no-one in mainstream AI academia is working on the control problem for superintelligence

• The_Jaded_One 29 Jan 2015 7:42 UTC

  1 point

  in reply to: JoshuaZ’s comment on: (mis­lead­ing ti­tle re­moved)

  The orthogonality thesis, in the form “you can’t get an ought from an is”, is widely accepted or at least widely considered a popular position in public discourse.

  It is true that slow superintelligence is less risky, but that argument isn’t explicitly made in this letter.

• The_Jaded_One 1 Mar 2015 10:45 UTC

  1 point

  in reply to: joaolkf’s comment on: An alarm­ing fact about the anti-ag­ing community

  God, I need an anti cryocrastination angel too!

• The_Jaded_One 1 Mar 2015 10:52 UTC

  0 points

  in reply to: Froolow’s comment on: An alarm­ing fact about the anti-ag­ing community

  Realistically, I would expect that having your own cells frozen probably brings more benefits on average than being able to get an organ donation. Organs are in short supply. If they were abundant, doctors would give them to more people, increasing your p given above.

  Still, if organs were free and 100% immunocompatible, many things would still be a problem. E.g. cancer.

Re­quest for help: An­droid app to shut down a smart­phone late at night

• The_Jaded_One 2 Apr 2015 12:39 UTC

  2 points

  in reply to: Sean_o_h’s comment on: Re­quest for help: An­droid app to shut down a smart­phone late at night

  I already have another phone. Yes, this is a good point though.