Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Comment author: cousin_it 21 June 2017 12:52:38PM *  2 points [-]

The argument somehow came to my mind yesterday, and I'm not sure it's true either. But do you really think human value might be as easy to maximize as pleasure or pain? Pain is only about internal states, and human value seems to be partly about external states, so it should be way more expensive.

Comment author: wallowinmaya 21 June 2017 06:41:50PM *  4 points [-]

One of the more crucial points, I think, is that positive utility is – for most humans – complex and its creation is conjunctive. Disutility, in contrast, is disjunctive. Consequently, the probability of creating the former is smaller than the latter – all else being equal (of course, all else is not equal).

In other words, the scenarios leading towards the creation of (large amounts of) positive human value are conjunctive: to create a highly positive future, we have to eliminate (or at least substantially reduce) physical pain and boredom and injustice and loneliness and inequality (at least certain forms of it) and death, etc. etc. etc. (You might argue that getting "FAI" and "CEV" right would accomplish all those things at once (true) but getting FAI and CEV right is, of course, a highly conjunctive task in itself.)

In contrast, disutility is much more easily created and essentially disjunctive. Many roads lead towards dystopia: sadistic programmers or failing AI safety wholesale (or "only" value-loading or extrapolating, or stable self-modification), or some totalitarian regime takes over, etc. etc.

It's also not a coincidence that even the most untalented writer with the most limited imagination can conjure up a convincing dystopian society. Envisioning a true utopia in concrete detail, on the other hand, is nigh impossible for most human minds.

Footnote 10 of the above mentioned s-risk-static makes a related point (emphasis mine):

"[...] human intuitions about what is valuable are often complex and fragile (Yudkowsky, 2011), taking up only a small area in the space of all possible values. In other words, the number of possible configurations of matter constituting anything we would value highly (under reflection) is arguably smaller than the number of possible configurations that constitute some sort of strong suffering or disvalue, making the incidental creation of the latter ceteris paribus more likely."

Consequently, UFAIs such as paperclippers are more likely to create large amounts of disutility than utility (factoring out acausal considerations) incidentally (e.g. because creating simulations is instrumentally useful for them).

Generally, I like how you put it in your comment here:

In terms of utility, the landscape of possible human-built superintelligences might look like a big flat plain (paperclippers and other things that kill everyone without fuss), with a tall sharp peak (FAI) surrounded by a pit that's astronomically deeper (many almost-FAIs and other designs that sound natural to humans). The pit needs to be compared to the peak, not the plain. If the pit is more likely, I'd rather have the plain.

Yeah. In a nutshell, supporting generic x-risk-reduction (which also reduces extinction risks) is in one's best interest, if and only if one's own normative trade-ratio of suffering vs. happiness is less suffering-focused than one's estimate of the ratio of expected future happiness to suffering (feel free to replace "happiness" with utility and "suffering" with disutility). If one is more pessimistic about the future or if one needs large amounts of happiness to trade-off small amounts of suffering, one should rather focus on s-risk-reduction instead. Of course, this simplistic analysis leaves out issues like cooperation with others, neglectedness, tractability, moral uncertainty, acausal considerations, etc.

Do you think that makes sense?

Comment author: cousin_it 20 June 2017 01:46:47PM *  10 points [-]

Wow!

Many thanks for posting that link. It's clearly the most important thing I've read on LW in a long time, I'd upvote it ten times if I could.

It seems like an s-risk outcome (even one that keeps some people happy) could be more than a million times worse than an x-risk outcome, while not being a million times more improbable, so focusing on s-risks is correct. The argument wasn't as clear to me before. Does anyone have good counterarguments? Why shouldn't we all focus on s-risk from now on?

(Unsong had a plot point where Peter Singer declared that the most important task for effective altruists was to destroy Hell. Big props to Scott for seeing it before the rest of us.)

Comment author: wallowinmaya 21 June 2017 06:50:26AM 2 points [-]

The article that introduced the term "s-risk" was shared on LessWrong in October 2016. The content of the article and the talk seem similar.

Did you simply not come across it or did the article just (catastrophically) fail to explain the concept of s-risks and its relevance?

Comment author: wallowinmaya 16 April 2017 04:33:20PM 1 point [-]

Here is another question that would be very interesting, IMO:

“For what value of X would you be indifferent about the choice between A) creating a utopia that lasts for one-hundred years and whose X inhabitants are all extremely happy, cultured, intelligent, fair, just, benevolent, etc. and lead rich, meaningful lives, and B) preventing one average human from being horribly tortured for one month?"

Comment author: wallowinmaya 14 April 2017 09:51:17AM *  1 point [-]

I think it's great that you're doing this survey!

I would like to suggest two possible questions about acausal thinking/superrationality:

1)

Newcomb’s problem: one box or two boxes?

  • Accept: two boxes
  • Lean toward: two boxes
  • Accept: one box
  • Lean toward: one box
  • Other

(This is the formulation used in the famous PhilPapers survey.)

2)

Would you cooperate or defect against other community members in a one-shot Prisoner’s Dilemma?

  • Definitely cooperate
  • Leaning toward: cooperate
  • Leaning toward: defect
  • Definitely defect
  • Other

I think that these questions are not only interesting in and of itself, but that they are also highly important for further research I'd like to conduct. (I can go into more detail if necessary.)

Comment author: wallowinmaya 09 April 2017 07:42:18AM *  14 points [-]

First of all, I don't think that morality is objective as I'm a proponent of moral anti-realism. That means that I don't believe that there is such a thing as "objective utility" that you could objectively measure.

But, to use your terms, I also believe that there currently exists more "disutility" than "utility" in the world. I'd formulating it this way: I think there exists more suffering (disutility, disvalue, etc.) than happiness (utility, value, etc.) in the world today. Note that this is just a consequence of my own personal values, in particular my "exchange rate" or "trade ratio" between happiness and suffering: I'm (roughly) utilitarian but I give more weight to suffering than to happiness. But this doesn't mean that there is "objectively" more disutility than utility in the world.

For example, I would not push a button that creates a city with 1000 extremely happy beings but where 10 people are being tortured. But a utilitarian with a more positive-leaning trade ratio might want to push the button because the happiness of the 1000 outweighs the suffering of the 10. Although we might disagree, neither of us is "wrong".

Similar reasoning applies with regards to the "expected value" of the future. Or to use a less confusing term: The ratio of expected happiness to suffering of the future. Crucially, this question has both an empirical as well as a normative component. The expected value (EV) of the future for a person will both depend on her normative trade ratio as well as her empirical beliefs about the future.

I want to emphasize, however, that even if one thinks that the EV of the future is negative, one should not try to destroy the world! There are many reasons for this so I'll just pick a few: First of all, it's extremely unlikely that you will succeed and will probably only cause more suffering in the process. Secondly, planetary biocide is one of the worst possible things one can do according to many value systems. I think it's extremely important to be nice to other value systems and promote cooperation among their proponents. If you attempted to implement planetary biocide you would cause distrust, probably violence and the breakdown of cooperation, which will only increase future suffering, hurting everyone in expectation.

Below, I list several more relevant essays that expand on what I've written here and which I can highly recommend. Most of these link to the Foundational Research Institute (FRI) which is not a coincidence as FRI's mission is to identify cooperative and effective strategies to reduce future suffering.

I. Regarding the empirical side of future suffering

II. On the benefits of cooperation

III. On ethics

Comment author: wallowinmaya 08 April 2017 05:18:11PM 0 points [-]

Great list!

IMO, one should add Prescriptions, Paradoxes, and Perversities to the list. Maybe to the section "Medicine, Therapy, and Human Enhancement".

[Link] Decision Theory and the Irrelevance of Impossible Outcomes

2 wallowinmaya 28 January 2017 10:16AM

[Link] Why Altruists Should Focus on Artificial Intelligence

1 wallowinmaya 16 December 2016 11:48AM
In response to Seven Apocalypses
Comment author: wallowinmaya 29 September 2016 04:30:29PM *  3 points [-]

I don't understand why you exclude risks of astronomical suffering ("hell apocalypses").

Below you claim that those risks are "Pascalian" but this seems wrong.

[Link] How the Simulation Argument Dampens Future Fanaticism

6 wallowinmaya 09 September 2016 01:17PM

Very comprehensive analysis by Brian Tomasik on whether (and to what extent) the simulation argument should change our altruistic priorities. He concludes that the possibility of ancestor simulations somewhat increases the comparative importance of short-term helping relative to focusing on shaping the "far future".

Another important takeaway: 

[...] rather than answering the question “Do I live in a simulation or not?,” a perhaps better way to think about it (in line with Stuart Armstrong's anthropic decision theory) is “Given that I’m deciding for all subjectively indistinguishable copies of myself, what fraction of my copies lives in a simulation and how many total copies are there?"

 

View more: Next