211–220 of 220 posts
Earlier quoted context omitted.
I can't think of one that avoids this problem. Can you?
I did throw out "probability/quality adjusted years of life gained" to address the obviously grossly inadequate metric of counting "lives gained/lost".
Earlier quoted context omitted.
I did throw out "probability/quality adjusted years of life gained" to address the obviously grossly inadequate metric of counting "lives gained/lost".
That can't really be it, can it? Then we just add patients who need a blood transfusion, two who need corneas, someone who needs a liver, and so on, until we're positive that the expected value of life is higher.
I kind of think that society does use something like the measure of utility I described - we just don't talk about it a lot. For instance, doctors will refuse to do an operation on someone who is sufficiently old and frail. Courts will award less to an elderly victim of medical error. And so on.
Edit/post script: Also, I think you implicitly are assuming that utility is one dimensional, and that you can add up quality of life changes and balance them against lives ended. That does not necessarily seem valid or essential to some form of utilitarianism.
Earlier quoted context omitted.
That can't really be it, can it? Then we just add patients who need a blood transfusion, two who need corneas, someone who needs a liver, and so on, until we're positive that the expected value of life is higher.
This seems to me kind of like saying if you pick an arbitrary number k, and an infinite sum from 1..n, there must be a point at which the sum is greater than k. But if the sum is finite, it may be less than k for any n. And if you're talking about real life, the total benefit is uncertain too. So I am not at all convinced that this kind of utilitarianism easily leads to decisions that we don't see in real life. I kin…
It seems that way to me. What difference does it make to utility whether a life ended naturally or unnaturally? Indeed, death from organ failure is probably much more painful than whatever I might do as an unethical doctor. Once you're looking at utility + non-utilitarian considerations like "murder is wrong," it's simply not utilitarianism anymore.
Since we're hung up on details, let's look at other examples:
* You're a judge in a frontier town with only a rudimentary justice system wherein the only possible punishment is hanging the accused person on the spot after they are found guilty. A townsperson is accused of witchcraft. You know this defendant is not guilty, because there is no such thing as witchcraft, but you are also certain that if you rule that the defendant is not guilty a riot will break out and several people will be killed. Therefore, the utility-maximizing option would seem to be falsely finding the defending guilty and ordering execution. Is that the ethical thing to do?
* You are preparing a body for burial. The body is to be buried with several pieces of precious jewelry, per the family's request. You know that if you steal the jewelry, replace it with costume jewelry, pawn the real jewelry, and use the funds to buy your friends dinner, nobody will be any the wiser, and you will have maximized utility by using the jewelry to bring your friends happiness rather than burying it. Ethical?
Earlier quoted context omitted.
This seems to me kind of like saying if you pick an arbitrary number k, and an infinite sum from 1..n, there must be a point at which the sum is greater than k. But if the sum is finite, it may be less than k for any n. And if you're talking about real life, the total benefit is uncertain too. So I am not at all convinced that this kind of utilitarianism easily leads to decisions that we don't see in real life. I kin…
> Edit/post script: Also, I think you implicitly are assuming that utility is one dimensional, and that you can add up quality of life changes and balance them against lives ended. That does not necessarily seem valid or essential to some form of utilitarianism. It seems that way to me. What difference does it make to utility whether a life ended naturally or unnaturally? Indeed, death from organ failure is probably…
Edit/post script: I am not so sure that rules like "don't murder" are in conflict with utilitarianism as I think of it existing. Maybe if you look at moral choices as an optimization process, simple rules function as heuristics to deal with uncertainty, bias and lack of information on utility.
Earlier quoted context omitted.
> Edit/post script: Also, I think you implicitly are assuming that utility is one dimensional, and that you can add up quality of life changes and balance them against lives ended. That does not necessarily seem valid or essential to some form of utilitarianism. It seems that way to me. What difference does it make to utility whether a life ended naturally or unnaturally? Indeed, death from organ failure is probably…
You seem to be hung up on utilitarianism implying a particular one dimensional utility metric. I don't see why that is necessary, to call something "utilitarianism". Why can't utilitarianism be conceived of as parameterizable? Without arguing for any particular ethical system, I think the world as it exists operates largely on a utilitarian basis, just with a less simplistic utility metric, similar to what I mentione…
Earlier quoted context omitted.
If something goes wrong with an entity dealing with over a hundred million dollars, the CEO is expected to respond relatively immediately, as in, on call 24/7, because a hundred million represents the interests of a lot of people most likely. The variety of things that can go wrong is large, and the necessary responses are nuanced, potentially requiring sophisticated knowledge in several domains, from legal, accounti…
For charities, a pertinent question is: Why should I donate money to someone who takes home more money than I do? Why isn't he donating?
If you're struggling, feed yourself first. Don't create new needs, robbing Peter to pay Paul. If you have something extra, then sit down and figure out where your dollars actually do the most good, and resist the temptation to donate randomly with no research, based on a heartfelt appeal or good promotional materials.
Maybe GiveWell doesn't meet that test for you, that's fine, if everyone just thought it through that much we'd be way better off than we are now.
I want to make sure it's said that effective altruism is intertwined with the rationalist movement (Scott Alexander[1], Eliezer Yudkowsky[2], Robin Hanson[3]) and that, among these so called rationalists and the parts of the effective altruist community where they hold sway, there are a lot of advocates of AI risk mitigation research (what this means, who knows). These people see AI as the greatest risk to humanity,…
First off, yes, there is overlap between effective altruism and the rationalism community. I think that makes sense when you want to try to use reason instead of intuition to make decisions.
I fail to see why you should qualify them as "so called" rationalists. Though besides Yudkowsky (and maybe even him) I honestly haven't heard enough of them to defend them. If you do want to sling in some defamatory remarks at their expense I feel like they should be backed up (or left unsaid), though.
You mention that some people see AI as the greatest risk to humanity. Perhaps I misunderstand but the way you phrase this, it sounds like you think this is absolutely ridiculous. If so, why would that be? And what long list of assumptions would you need to agree to? And why would it be bad for there to be a common starting point for discussing this? I think there's a fair amount of uncertainty in how any superintelligent entity would act, so a certainty of AI being terrible seems silly. However, a strong belief that it can pose a large threat seems, honestly, evident.
You say people use "AI risk" as a front to recruit people into their belief system. There is so much wrong with this... first off, why is it a front? That implies deception. Secondly, it is one of many facets. The core of EA is the desire to do have a (large) positive impact. If some people think that they can make their impact by working on the AI safety issue, why do you feel the need to portray that as nefarious? Finally, "belief system" sounds incredibly dogmatic. EA is not a church. Yes, there is a set of beliefs that most people in the EA community would ascribe to. But I don't experience EA as some echo chamber where everyone is forced into some kind of mold. Rather, people challenge both each other's and their own ideas. There's inevitably going to be some biases and filters, but your portrayal of EA as a cult (purposefully or not) is inaccurate.
As far as EA communities tending to be alt-right... what on earth are you smoking? I help run a local chapter and the focus is highly left-wing. And anyone I've noticed that's slightly more right wing is definitely not of the misoginistic or racist side. I recently listened to an 80000 hours podcast with Bryan Caplan, and noticed he's libertarian. While I think libertarian views are mostly bonkers, at the very least the way his libertarian views showed (e.g. arguing for open borders) are not insane on the level I'm used to from libertarians. Either way, this is an exception. Even if you can list some well known names that also have some strange views, I can say with a high degree of certainty that it is not even remotely representative of the community as a whole, especially not as I've seen it in the Netherlands.
FINALLY: I've honestly only ever seen Roko's Basilisk being mentioned on a meme page for EA. So much for taking it seriously.
Earlier quoted context omitted.
That's a fair point, but ultimately people are going to care more about children that are 'closer' to them. You're free to criticize people for this, but it probably won't do you any favours.
> That's a fair point, but ultimately people are going to care more about children that are 'closer' to them. You're free to criticize people for this, but it probably won't do you any favours. Also, making charity too much about distant problems and far off research outcomes factors out the civic and community-level thinking that actually drives and sustains a lot of charity. It also makes charity solely the domain…
It makes sense not to feel morally responsible for things that happen out of view when you cannot know what is going on and have no way to impact it. Thing is, we do know a lot about what is going on and we do have tools to impact it. It just doesn't feel intuitively satisfying. It is possible to internalize this satisfaction regardless, though.
Earlier quoted context omitted.
Not remotely applicable. For one, so far everyone predicting doom about AI has been a layman a subject. Maybe you aren't aware, but this is a field that people do PhDs and get professorships in. Noone respected agrees with the doom-sayers.
> Maybe you aren't aware, but this is a field that people do PhDs and get professorships in. I'm aware of my own existence, thanks ;-) > Noone respected agrees with the doom-sayers But many, many, many respected AI researchers (who I've talked to about this) certainly do agree that AI/ML safety/robustness are important topics of inquiry. Including researchers at all the top CS departments, at Deepmind, at OpenAI, etc…
> Consider critiquing the concrete projects that are funded with AI safety money. Very little of that money flows to research about "preventing malicious superintelligence" or whatever strawman you have in mind. And even projects that consider those questions also consider many much more near-term AI safety questions.
But this ("preventing malicious superintelligence") is EXACTLY what we are talking about. The comment in the OP was quoting the following
> Grants will likely go to organizations that seek to reduce global catastrophic risks, especially those relating to advanced artificial intelligence.