following on from part of the discussion in the unethical vegetables thread,

it seems some utilitarians expect that future humans will create a massive amount of happy sentients, astronimcal waste argument, etc

and then others are worried that wildlife might be spread to other planets or simulated or many other things and that we might have astronomical amounts of suffering instead

what probabilities do you think future outcomes have and why. and how do you think we should act on them?

Hi Ruairi! A very interesting issue!

I guess future won't be a black or white scenario. Perhaps there will be both happy sentient beings and unhappy sentient beings in long-term future.

If there are humans (or post-humans) in long-term future, I'd put the odds around 75% (a worse universe than now) and 25% (a better universe than now). My odds are wildly intuitive, but I guess them when I see how humans are used technology for the last years, producing a lot of suffering to animals.

I think the best we can do is working for a better future world. So I try there are more people giving moral considerability to (domestic and wild) animals. This increases the chances of a better world, and I guess it's more cost-effective than working against (or for) existential risk.

The future universe has the potential to be be very good (creating a lot of happiness) or very bad (creating a lot of suffering).

Considering that human civilisation has improved so far (cf Pinker's history of violence), and thinking that this could continue into the future, maybe we have an overwhelming obligation to preserve human civilisation. It seems the potential happiness that our civilisation can bring about far outweighs present concerns such as poverty and animal cruelty. There here are a couple of cautions for this approach:
> The accuracy of our predictions decreases (?exponentially) over time. We have to discount future potential happiness according to its uncertainty.
> For humanity to have positive utility, it would possibly have to be more altruistic than another alien species that would occupy our part of the universe.

Furthermore, even if decreasing existential risk is the overwhelming obligation, this can be done in various ways, such as by direct research, funding, or evangelising the issue. Furthermore, actually solving the world's other problems is a way of evangelising the issue. If all humans on Earth flourished and no animals suffered, organisations that work to reduce existential risk would recieve much greater funding. The case for SIAI would be more straightforward: the world is great, let us not risk what we have! Very persuasive, right? Then maybe we should address poverty and animal suffering in order to preserve human civilisation from existential risk. Whether or not this is a useful approach is a matter for science.

Thanks, Ruairi. This seems to be the eternal question on this forum.

Here are past threads on the topic:

"Are increases in existential risks good or bad?"
"A few dystopic future scenarios"
"Friendly AI and utilitarianism"

I think the net balance of happiness vs. suffering on earth is currently negative, in light of the vast numbers of small, short-lived organisms that die a few days after birth (not to mention more conventional concerns like torture, genocide, famine, mental illness, etc.). A post-human future might mean creation of lots more beings that have experiences like these (through panspermia, lab universes, and simulations), and to this extent, the net-negative balance of suffering would multiply many-fold.

I haven't really thought about this issue enough to have good numerical probability estimates, but let me throw something out there so that it can be criticized and refined.

P(we create vastly more happiness than suffering | humans survive and become post-humans) = 17%
P(we create some more happiness than suffering | humans survive and become post-humans) = 5%
P(we create about the same amount of happiness as suffering | humans survive and become post-humans) = 25%
P(we create some more suffering than happiness | humans survive and become post-humans) = 40%
P(we create vastly more suffering than happiness | humans survive and become post-humans) = 13%.

In addition to trying to change the probability of human survival, we can also make a difference by trying to change the outcomes that happen if humans do survive. This is my preferred target for action, because I'm currently dubious that human survival has net-positive expected value (especially from my negative-utilitarian-leaning standpoint), and the number of people who are working to prevent existential risk is comparatively much larger than the number working to spread concern for wild animals.

cool! thanks so much for all the replies!:D! ill read all the links and stuff soon thanks:)!

just some things i wanted to throw out there:

whats the probability of humans spreading vast amounts of life and then going extinct?

the probability we wont create massive numbers of sentients in the future?

is there anyway we can predict the future better? any useful research that could be done?

Ruairi wrote:whats the probability of humans spreading vast amounts of life and then going extinct?

the probability we wont create massive numbers of sentients in the future?

Here are again some very intuitive, unmeditated numbers. I haven't factored anthropics (doomsday argument, simulation argument, etc.) into this either.

P(humans survive and develop AGI) = 30%
P(directed panspermia | humans survive and develop AGI) = 12%
P(terraforming | humans survive and develop AGI) = 15%
P(lab universes | humans survive and develop AGI) = 0.001%
P(a few sentient simulations | humans survive and develop AGI) = 55%
P(massive numbers of sentient simulations | humans survive and develop AGI) = 20%.

If we account for the simulation argument, all of these values decrease quite a bit, in rough proportion to the amount of computing resources they require. Of course, the same will be true for probabilities of happy simulations.

Ruairi wrote:is there anyway we can predict the future better? any useful research that could be done?

SIAI has a page called The Uncertain Future, although it's primarily about probabilities of AGI alone, without considering the many possible social implications thereof. There are lots of futurology publications, and even if their estimates are way off, they might suggest new ideas you hadn't thought of. For that matter, even science fiction can be a great way to stretch the imagination, but keep in mind that most science fiction is highly improbable and prone to good-story bias. I think a paperclipping future is fairly likely, but it doesn't make for a very interesting story.

There's a saying, "The best way to predict the future is to create it." While this is obviously a hugely false hyperbole, it hints at what I was suggesting in my previous post: That regardless of whether the future is net good or bad, we can still push it in a slightly better direction by raising concern for animal suffering, etc.

cool thanks!:D! one last thing i was wondering was if humans were to go extinct tomorrow how long till something kills all sentient life/all life on earth? cause if this is really long then thats a lot of suffering

Ruairi wrote:one last thing i was wondering was if humans were to go extinct tomorrow how long till something kills all sentient life/all life on earth?

I'd say roughly one billion years, maybe more for small very resilient organisms. There's also the possible number of other planets with suffering that a space-colonizing process could reach and "rescue". But it still seems quite clear to me that this amount is significantly lower than the amount of suffering a space-colonizing system would create if it retained any suffering in its process, due to the gigantonormic negentropy resources that hi-tech could turn into consciousness. Same for lab universes etc.

I noticed with interest that David Pearce has advocated active resource investment in reducing existential risk in this interview from last October (see question 9). So either he's super confident about abolitionism + the need for cosmic rescue missions, or he's reasoning from a total rather than negative utilitarian logic.

Hedonic Treader wrote:I'd say roughly one billion years

Yes, I've heard the same estimate, plus or minus 0.5 billion years, though I can't find the source paper at the moment. So we're actually more than half-way through the lifetime of terrestrial biology.

Hedonic Treader wrote:There's also the possible number of other planets with suffering that a space-colonizing process could reach and "rescue". But it still seems quite clear to me that this amount is significantly lower than the amount of suffering a space-colonizing system would create if it retained any suffering in its process, due to the gigantonormic negentropy resources that hi-tech could turn into consciousness. Same for lab universes etc.

I was going to say exactly the same thing. There might be a few other planets with suffering life, or with civilizations intent on creating lab universes, but getting to them physically would be hard, and there probably aren't that many within reach. On balance we're probably more likely to spread new suffering life than to rescue existing suffering life.

In addition to trying to change the probability of human survival, we can also make a difference by trying to change the outcomes that happen if humans do survive. This is my preferred target for action, because I'm currently dubious that human survival has net-positive expected value (especially from my negative-utilitarian-leaning standpoint), and the number of people who are working to prevent existential risk is comparatively much larger than the number working to spread concern for wild animals.

Brian, what are your views about our ability to reduce the probability of human extinction vis-à-vis our ability to comparably reduce the suffering experienced by posterity if humans do survive? It seems to me that it should be easier to prevent extinction than to benefit posterity. Extinction risks are concentrated on a very small fraction of the total period of possible Earth-originating sentient life (a few centuries or millennia in a total period of billions of years). Moreover, this risk is temporally located very close to us. By contrast, future sentients will likely be distributed more-or-less evenly over that astronomical period (assuming instead a continuously growing population actually strengthens my conclusion, since relatively more sentients will then be located in the farther future). On the plausible assumption that it is easier for us to have a deliberate causal influence over what is closer to us than over what is temporally removed, it seems we must conclude that affecting the risk of human extinction should be much easier than affecting the welfare of posthuman sentients to a comparable degree. Do you agree with this assessment? Or are you assuming some kind of mechanism whereby changes to the welfare of the first few generations of posthumans will be maintained indefinitely over time (this would be the case if the welfare of a given generation completely determines the welfare of the next generation)? Or perhaps you have other considerations in mind?

Thanks, Pablo!

Pablo Stafforini wrote:Do you agree with this assessment?

You make a good case. The problem is, I don't know if reducing extinction risk is a good idea. I'm also not willing to work to increase that risk, so to some extent, changing society's values is my only option. See also the next reply.

Pablo Stafforini wrote:Or are you assuming some kind of mechanism whereby changes to the welfare of the first few generations of posthumans will be maintained indefinitely over time (this would be the case if the welfare of a given generation completely determines the welfare of the next generation)? Or perhaps you have other considerations in mind?

It's not the direct welfare of posthumans that I'm interested in but, rather, their moral sentiments about creating vast numbers of new wild animals and suffering subroutines. Preservation of ethical values is difficult, but that's exactly one of the things SIAI is working on, so at least in the case where SIAI succeeded in that effort (unlikely as it may be), then changing humanity's views would have consequences into the indefinite future. Even if goal preservation isn't perfect, we can achieve some amount of shift in the right direction, although as you say, it's unclear how much it would get corrupted as time went on.

Alan Dawrst wrote:It's not the direct welfare of posthumans that I'm interested in but, rather, their moral sentiments about creating vast numbers of new wild animals and suffering subroutines.

Yes, that was what I meant. I now realize it was misleading to use the term 'posthumans' to designate all Earth-originating future sentients. Unfortunately there doesn't seem to be any other term that picks out that class of beings.

Alan Dawrst wrote:Preservation of ethical values is difficult, but that's exactly one of the things SIAI is working on, so at least in the case where SIAI succeeded in that effort (unlikely as it may be), then changing humanity's views would have consequences into the indefinite future. Even if goal preservation isn't perfect, we can achieve some amount of shift in the right direction, although as you say, it's unclear how much it would get corrupted as time went on.

Thanks for the clarification!

RyanCarey wrote:> For humanity to have positive utility, it would possibly have to be more altruistic than another alien species that would occupy our part of the universe.

I don't understand what you mean here? Is this from an averaging perspective?

how could we go about researching this question of how the future looks? Or is it simply unlikely to really yield much return in information no matter how much we look into it?