Yeah, but assuming your p(doom) isn’t really high, this needs to balanced against the chance that AI goes well, and your kid has a really, really, really good life.
I don’t expect my daughter to ever have a job, but think that in more than half of worlds that seem possible to me right now, she has a very satisfying life—one that is better than it would be otherwise in part because she never has a job.
Timothy Underwood
Karma: 420
I’d note that acoup’s model of fires primacy making defence untenable between hi tech nations, while not completely disproven by the Ukraine war, is a hypothesis that seems much less likely to be true/ less true than it did in early 2022. The Ukraine war has shown in most cases a strong advantage to a prepared defender and the difficulty of taking urban environments.
The current Israel—Hamas was shows a similar tendency, where Israel is moving very slowly into the core urban concentrations (ie it has surrounded Gaza city so far, but not really entered it), though its superiority in resources relative to its opponent is vastly greater than Russia’s advantage over Ukraine was.
I’d expect per capita war deaths to have nothing to do with offence/ defence balance as such (unless the defence gets so strong that wars simply don’t happen, in which case it goes to zero).
Per capita war deaths in this context are about the ability of states to mobilize populations, and about how much damage the warfare does to the civilian population that the battle occurs over. I don’t think there is any uncomplicated connection between that and something like ‘how much bigger does your army need to be for you to be able to successfully win against a defender who has had time to get ready’.
This matches my sense of how a lot of people seem to have… noticed that GPT-4 is fairly well aligned to what the OpenAI team wants it to be, in ways that Yudkowsky et al said would be very hard, and still not view this as at a minimum a positive sign?
Ie problems of the class ‘I told the intelligence to get my mother out of the burning building and it blew her up so the dead body flew out the window, this is because I wasn’t actually specific enough’ just don’t seem like they are a major worry anymore?Usually when GPT-4 doesn’t understand what I’m asking, I wouldn’t be surprised if a human was confused also.
Weirdly, and I think this is because my childhood definitely was not optimized for getting into a good university (I was homeschooled, and ended up transferring to Berkley based off perfect grades for two years in a community college), but reading the last paragraphs here made me rather nostolgic for the two or three weeks I spent doing practice SAT tests.
I mean, it kind of does fine at arithmetic?
I just gave gpt3.5 three random x plus y questions, and it managed one that I didn’t want to bother doing in my head.
I think the issue is that creating an incentive system where people are rewarded for being good at an artificial game that has very little connection to their real world cericumstances, isn’t going to tell us anything very interesting about how rational people are in the real world, under their real constraints.
I have a friend who for a while was very enthused about calibration training, and at one point he even got a group of us from the local meetup + phil hazeldon to do a group exercise using a program he wrote to score our calibration on numeric questions drawn from wikipedia. The thing is that while I learned from this to be way less confident about my guesses—which improves rationality, it is actually, for the reasons specified, useless to create 90% confidence intervals about making important real world decisions.
Should I try training for a new career? The true 90% confidence interval on any difficult to pursue idea that I am seriously considering almost certainly includes ‘you won’t succeed, and the time you spend will be a complete waste’ and ‘you’ll do really well, and it will seem like an awesome decision in retrospect’.
If you think P(doom) is 1, you probably don’t believe that terrorist bombing of anything will do enough damage to be useful. That is probably one of EYs cruxes on violence.
You don’t become generally viewed by society as a defector when you file a lawsuit. Private violence defines you in that way, and thus marks you as an enemy of ethical cooperators, which is unlikely to be a good long term strategy.
Yeah, but I read somewhere that loneliness kills. So actually risking being murdered by grass is safer, because you’ll be less lonely.
I think we agree though.
Making decisions based on tiny probabilities is generally a bad approach. Also, there is no option that is actually safe.
You are right that I have no idea about whether near complete isolation has a higher life expectancy than being normally social, and the claim needed to compare them to make logical sense in that way.
I think the claim does still make sense if interpreted as ‘whether it is positive or negative on net, deciding to be completely isolated has way bigger consequences, even in terms of direct mortality risk, than taking the covid vaccine’ - and thus avoiding the vaccine should not be seen as a major advantage of being isolated.
My experience is that it is like having extra in laws, who you may or may not like, but have to sort of get along with occasionally.
I don’t think most people actually talk very much with their in laws, or assume that people who the in law dislikes should be disliked.
What I meant is that it is possible the things that Von Neumann discovered were easier to discover than anything that is still undiscovered, so new Von Neumann’s won’t be as impressive.
“this is something that the data has to actually exist for since several percent of US children have been homeschooled for the last several decades.”
Never mind. There aren’t particularly good studies. But what exists seems to say that homeschooled students do much better than average for all students, but maybe somewhat worse than the average for students with their parent’s SES backgrounds.
But the data mostly comes from non-random samples, so it is hard to generate firm conclusions.
So this is based on my memory of homeschooling propaganda articles that I saw as a kid. But I’m pretty sure the data they had there showed most kids went to college. In my family three of us got University of California degrees, and the one who only got a nursing degree in his thirties authentically enjoyed manual labor jobs until he decided he also wanted more money.
Perhaps these numbers do stop at college, and so we don’t see in them children who get a good college education, but then fail in some important way later on in life, but I’ve never gotten an impression from anywhere that homeschooled children have generally worse life outcomes—anyways, this is something that the data has to actually exist for since several percent of US children have been homeschooled for the last several decades.
I did have substantial social problems, even as an adult, and they have led me to be less successful in career terms than I probably would have been with stronger social skills. But this might be driven by a selection effect: The reason my parents actually started homeschooling me was because I was being bullied and having severe social problems in third grade.
oops, that was supposed to be something like ‘low hanging fruit’, I’m pretty sure it was a typo.
I recently looked through the wikipedia list of the thirty richest Americans, and then tried to dig back into their class background (or the class background of the founder of the family fortune for heirs, like the Walton family). In almost every single case where I could identify the class background, they were from a top couple of percent background, but in only a few cases were they from an old money background. So a lot of the founders of big fortunes have backgrounds like ‘father was a lawyer/ stockbroker/ ran a grocery store/ dentist/ college professor/ middle manager’.
One interesting feature here was that there were several Russian immigrants or children of immigrants on the list (usually they moved to the US before they were a teenager, and usually they were Jewish). In these cases I found that I have generally no idea what class status is implied by the descriptions of their parent’s work in the Soviet Union. But I sort of suspect it usually was still top couple of percent.
I then looked at the European numbers, which were an interesting constrast in that:
A) A lot of the European super fortunes start with people who are as rich as far back as wikipedia tracks it. Ie the founder of the company got his money from his rich textile factory father (who doesn’t have a wikipedia account) in the late nineteenth century.
B) Weirdly, there were also more actual rags to riches stories in among the European superrich. The Zara founder is the one that stuck in my head. He seems to have been from definitely a lower two thirds of the income distribution household, and possibly even genuinely poor family in early Franco era Spain. There were several other stories that felt very much like ‘person with a totally normal family background somehow builds a giant fortune’, while again that seemed to not happen in the US listings.
I probably should make a post based on this at some point.
Or von Neumann and his contemporaries and predecessors stole all the insights that someone with merely Neumann’s intellect could develop independently, leaving future geniuses to have to be part of collaborative teams?
What specifically do you think is really high variance as opposed to the main downside being that it is expensive? If it is the ‘not going to school thing’ at least when I was growing up as a religiously homeschooled kid in the 90s, the strong impression that I got was that homeschooled kids systematically did better than other kids in terms of college success and other legible metrics—of course this has a gargantuan selection bias going on. But that does give a strong lower bound for how bad that specifically can be for kids.
The other stuff I recall from the article (ie being from a high resource background, having an intellectual mentor, being surrounded by intellectual conversations, getting one on one tutoring, good intrinsic capability) all seem to be things that either you can’t pick whether a child has or not, or where it would be weird if they left the child worse off.One on one tutoring, for example, just doesn’t seem like a high variance thing, it seems like a positive expected value thing that might not actually be that causally important or have that big of impact, but where it will only make things worse in exceptional cases.
Prime age labor force participation rate is the standard measure the econobloggers I’ve followed (most notably Krugman and Brad DeLong, who are part of the community also pushing for this interpretation of monetary policy) tend to use to measure economic health, and there are reasons to see it as pointing most closely to what we actually care about (that and hourly productivity, which isn’t in these charts).
You might capture value out of that relative to broad equities if the world ends up both severely deflationary due to falling costs, and where current publicly traded companies are mostly unable to compete in the new context.