Posts
Comments
I was actually taking allergy pills because they help with my asthma symptoms, and nasal sprays don't seem to help me. I started taking Singulair / Montelukast about halfway through when I was taking Zyrtec every day, and it seems to be more effective without the sleep-related issues (although the combination was still more effective for my asthma than either alone).
But yeah, for normal allergy symptoms, focusing the medicine on your nose instead of your entire blood stream is also a good idea. Theoretically there's even an antihistamine nose spray now (Astepro) although I haven't tried it.
I don't see anything about Buddhist monks or stop lights in the original article? I think you might be doing the thing where there's an argument for this inside of your head but you're not providing it to us.
It seems like your meanings are just based on Western culture. I would have expected a post about universal meanings to show how cultures pre-Western-influence used the same colors in the same ways.
It also seems like you haven't provided evidence for most of your meanings?
You're putting a lot of importance on the "likely" here:
Superintelligence is likely to arrive within the next 20 years
What if it doesn't? Reaching retirement age with no savings sucks pretty badly (especially if you're used to spending your entire income). Short AI timelines might push you in the direction of smaller retirement savings (taking a risk of things being a little worse in timelines you think are unlikely), but probably not all the way to "don't save for retirement". You should also put at least some weight on AGI happening and somehow the world still existing in a recognizable form (since this is what has happened every other time a world-changing technology was created).
Something else to consider is that retirement isn't the only reason to have savings. If you think AI timelines are short, you might want to have a giant pile of money you can strategically deploy to do things like quit your job and work on alignment for a few years (or pay someone else to).
Could you give more info about how and when you use this? Do you just use the sleep setting and the wakeup settings, or do you also use it during the day?
What intensity did you find worked well for you starting and what did that go down to now?
I want to try this out but I'm not sure how to replicate what you're doing
I already emailed you about this but it might be useful to share here too for feedback.
I think it would be bad to use a technology if:
- Using the technology itself is an existential risk (i.e. if ChatGPT might kill all humans if I ask the wrong question)
- Using it will increase the amount of resources into making it powerful enough for situation (1) to apply
I don't think (1) is currently the case with ChatGPT, and I think people like Eliezer agree.
I think using ChatGPT, even the paid version, won't increase the resources going into AI capabilities because the big AI companies aren't funding constrained. If OpenAI needed another billion dollars, they'd just sell a billion dollars worth of stock. My $20 per month probably increases the price people would pay for that stock, but that just reduces the dilution existing shareholders face and has no real effect on their ability to get funding if they need it.
I might feel differently if my usage increased other people's usage of ChatGPT (although I also think hype is so high it would be very difficult to meaningfully increase it), if I was part of a big enough coalition that our boycott was meaningful and noticable, or if I was using it at a scale where the money paid to OpenAI was significant (I would consider free-riding with open source models to avoid funding capabilities).
I am trying to think of something that's banned purely for having existential risk and coming up blank.
Weren't CFC's banned for existential reasons (although only after an alternative was found, because it would be better to die than not have refrigerators..)?
Isn't doing well with technological unemployment really easy if you have a good job now, through the magic of investing? The hard part would be figuring out what to do if you're currently in a low-paying job, but I doubt that's common here.
Accidental Terraforming is neutral, an effect that human civilization has had upon the planet of our birth. It isn’t necessarily good or bad, but it does require careful thought and inquiry to determine if this is the course we want to set.
I don't think this is neutral though. I agree that the Earth's feelings about the matter are irrelevant, but terraforming means making a planet more like Earth / more habitable for humans, but climate change is making the planet less like the Earth we're used to and plausibly less habitable for humans. Something like Venusforming makes it more clear that it's pushing the climate in a direction we may not want to go.
It's possible I'm just too tired, but I don't follow this argument. I think this is the section I'm confused about:
An AI that responds to "what's your name" with "I am Wintermute, destroyer of worlds" will have a psychology which necessarily creates that response in that circumstance, even if it's experience of saying it or choosing to say it isn't the same as a human's would be.
I'm putting all this groundwork down because I think there's a perception that RLHF is a limited tool, because the underlying artificial mind is just learning how to 'pretend to be nice,' but in this paradigm, we can see that the change in behavior is necessarily accompanied by a change in psychology.
Its mind actually travels down different paths than it did before, and this is sufficient to not destroy the world.
What do you mean by psychology here? It seems obvious that if you change the model to generate different outputs, the model will be different, but I'm not understanding what you mean by the model's psychology changing, or why we would expect that change to automatically make it safe?
I also think (4) is a big piece of this. GPT-4 isn't that good at programming, but it's shockingly good for someone with no long-term memory doing everything in their head with limited or no ability to self-reflect.
Am I wrong if my takeaway from the Nation Merit / Commended Student award issue is that selective school admissions are so insane that being in the top 3% of test-takers is worthless?
There are a lot of posts on LessWrong that aren't about x-risk at all so I don't see why this would be a problem.
Burning biofuel may help accelerate things in the short term, but it will always fall short of long-term sustainability.
Yeah, but you might as well take the short-term boost from burning the biosphere and then put solar panels on top.
I was only talking about funds given to the elderly. The disability system is very bad and I'd recommend not making that the inspiration or a UBI.
The comment seems to be saying that they will remove off-topic comments or low-effort posts on things that have been discussed endlessly here, not block posts about AI risk in general. It's fair to write posts about why you think AI risk is overblown and it's important for the community to have outside input, but also it's important to be able to write posts that aren't about re-arguing the same thing over and over or the community will atrophy and die.
Note that this anti-doom post has a reasonably high karma score for being a link post, presumably because the writer is actually aware of and engages the best arguments against her position.
I think the intention of Social Security is to provide enough money for you to survive, but not neccesarily live comfortably, which matches how UBI's are marketed. It's possible with inflation it's not even good enough for that, but I suspect if it actually caused major problems it would be fixed, since it's a very popular program.
I don't think solvency is a real problem for Social Security necessarily. For historical reasons we pretend that workers pay in and then get benefits and Social Security supports itself, but that's not true and hasn't been for a while, but the program is still popular. I expect that if Social Security's surplus runs out, the funding from general taxes / printing money will increase to compensate.
Note that really the only argument I've heard against Social Security is that it's too expensive, but that argument will have much less power if we're living in a world of such extreme abundance that most people don't need to work.
I'm not sure exactly how Australia does this, but in the US, the largest welfare program we have manages to be fairly universal, politically popular and not abusive (as far as I know). That program is Social Security / welfare for older people.
I think some important pieces of how it manages to work are:
- The requirements are very minimal so almost everyone who lives long[1] enough qualifies (11 years of earnings[2] and reaching the age of 62)
- There's no requirements going forward, so there's no point to re-checking (i.e. if I'm 62 this year, I'll definitely be 62-or-older next year)
- Older people are a big enough portion of the voting population that making this worse would be politically unfeasible
Part of the popularity is that there's a general agreement that people over a certain age should get a break, but people used to work basically until they died and the idea that 62 year olds should be given retirement is fairly new. I suspect if we slowly pushed the retirement age down, the age at which people consider retirement reasonable will also go down. Past a certain point you might reach some political instability when people in their 20's realize how much they're getting screwed over and insist that the system covers them too, and I think you might reach an equilibrium where everyone over age X gets benefits because they "deserve" it and everyone under age X gets benefits because it would be "unfair" to leave them out when they're the ones doing all of the actual work.
But basically, I think you'd get better results from trying to expand a popular program than trying to expand an unpopular program. Starting from old-age down might also be more efficient, as younger people adapt to new technology faster and so are less impacted by technology making them obsolete.
[1] And importantly, it feels universal since everyone expects to live long enough to collect on it [2] Or there's various ways to get social security benefits through a spouse even after divorce
Is this saying most Contra dances require an N95 mask and booster? Or is it an "or" thing?
I watched the video on https://www.charachorder.com/products/charachorder-one (the one with the kid with glasses). They type somewhat fast but it didn't really seem that much faster than I type with a QWERTY keyboard.
I noticed while typing this comment that I didn't type quite as fast as them, but mostly because I was trying to compose what I wanted to write, not because I couldn't type faster.
That's a good point. At least in the modern world the childcare aspect of school is an economic benefit for parents. It's worth pointing out that raising younger children was one of the useful jobs older children have historically done.
I think this source answers your questions better than I've been able to: https://eh.net/encyclopedia/child-labor-in-the-united-states/
The article is about historical child labor in the 1800's, and finds that children cost more than they produced, but they were able to do useful farm work past the age of 7.
He finds that children under 7 reduced the value of farm output, presumably because they reduced their mothers’ economic activities. For each child aged 7 to 12 the family’s output increased by about $16 per year – only 7 percent of the income produced by a typical adult male. Teen-aged females boosted family farm income by only about $22, while teen-aged males boosted income by $58.
It's unclear to me what the breakdown here is but children "age 15 and under" were a large portion of the manufacturing workforce.
In 1820 children aged 15 and under made up 23 percent of the manufacturing labor force of the industrializing Northeast. They were especially common in textiles, constituting 50 percent of the work force in cotton mills with 16 or more employees, as well as 41 percent of workers in wool mills, and 24 percent in paper mills.
Some options I can think of:
- Optimize your researcher's single-threaded performance by offloading as many unnecessary tasks as possible to different
coresworkers. For example, mow the researcher's lawn for them, cook for them, etc. - Speed up any learning parts by providing experts (i.e. if you want to know "Why is X?", find an expert on X to answer questions about it instead of needing your researcher to track down the answer from less-personalized sources). Or
spin off worker threadshave assistants fetch and correlate personalized descriptions even if they're not experts.
It's also possible that some of this work can be sped up by putting researchers working on the problems in contact with each other. My understanding is that it's generally more effective for people to bounce their ideas off of other smart people than to work alone.
I was just trying to respond to the original question. Being required to send kids to school raises the cost of kids because you can't have them do work during that time. Legal limits on working ages do the same thing, although officially-working isn't the only open (consider chores and family businesses). Other reasons school increases the cost of having kids is that parents may need to use their own time to transport kids to school and back, and to help with school work; and they may need to pay for school supplies.
Whether you think this is worth it depends on if you think the value kids get from going to school exceeds the costs of fewer people having kids / the costs to parents. My understanding is that Zvi thinks the value of school to kids is negative, but if you disagree then it wouldn't be surprising if you disagree on this cost/benefit tradeoff.
Something to be aware of with votes is that I (and I think many other people) vote based on where we think the total karma should be, not necessarily just "this is good" or "this is bad". I actually voted this up even though I think it's only ok, because I think the post is made me think and doesn't deserve negative karma even though I think it's wrong, but I would have voted it down if it had a very positive karma score.
I'm confused about the section about pleasure. Isn't the problem with paperclip maximizers that if they're capable of feeling pleasure at all, they'll feel it only while making paperclips?
It seems like the important thing to ask isn't whether we can know for certain if a person without a cerebellum is concious (since we don't really know that for non-self people in general). More interesting to me is the question of if we've asked someone without a cerebellum if they feel concious. If the answer is "Yes, and they said they're concious", then I think that's about as much evidence as we're likely to get.
I think the claim is that a ban would give an advantage to stealth defection because the stealth defector would work faster than people who can't work at all, while a regulation requiring open sharing of research would make stealth defection a disadvantage because the stealth defector has to work alone and in secret while everyone else collaborates openly.
I think it depends, since you could have a situation where a stealth defector knows something secret and can combine it with other people's public research, but it would also be hard for someone to get ahead in the first place while working alone/in secret.
Downvoting not because I disagree but because I think starting an argument about gun control is distracting in this context.
The Sydney part at the end is confusing to me. I thought GPT's don't have long-term memory / anything to checkpoint?
Thanks, this was really well written and I learned some things. Now I want to try writing my own adaptive Huffman coding algorithm..
I feel like the mistake the post is pointing out is that people think way too many things are games of chicken, and end up removing poor people's steering wheels "for their own good".
I think there really is a general rule that more options is better, and while there are exceptions, they deserve extra scrutiny, especially when making decisions for other people. There's a reason cars generally come with a steering wheel.
But would you trust the methods if I had checked, or would you still want to check it yourself?
No I didn't estimate the effect myself. I don't think that's a reasonable bar for commenting in this context. I also doubt it would make a difference if I had, since as a non-expert, my opinion of any given study's methods is probably not going to convince anyone.
Why did you decide that this isolated demand for rigor belongs on my comment and not the parent, or dozens of other comments?
Ah that makes sense. It still feels like the mechanism here is that you reduce house prices by making people poor though (you're reducing the buying power of these people by $500 and relying on side effects to also reduce house prices).
This argument ignores why the price (hypothetically) goes down. The price only goes down in this situation because you just made some people homeless! If your people who were paying for rent with sex all pay for rent with money instead (and thus don't become homeless), the price stays the same.
These arguments seem to come from misunderstandings of why prices changes.
Suppose we allow selling sex for rent. The number of rentable apartmants stays the same; however, there will more demand for them, because some people can now pay for them by non-monetary means. Because of this, the rent prices will increase, and that would just accelerate the rent-problem.
This is written in a way that hides the important fact that rents rise in this situation because more people have housing, and more people having housing is good.
You can write a fully-general argument against anything good like this. "Suppose we give poor people food. The amount of food stays the same, however, there will be more demand for it because some people don't have to pay for it. Because of this, food prices increase, and that will just accelerate the problem."
The problem of high prices is that some people can't get the thing, but if prices are increasing because more people can get the thing then pointing at the higher price as bad doesn't make any sense.
While exchanging kidneys for medical treatment is OK for me, it should not be mixed with the standard money market. The forces of money markets usually optimize for dollar value, which could be decoupled from human wellbeing. The result would be a worse state for everyone.
I'm confused about what this even means but isn't this a fully general argument against money and/or trade in general?
Do you have evidence of this happening? Poor people can already sell platelets and I've never heard someone say "This person doesn't deserve help because they haven't sold platelets," even though platelet sale is much easier than kidney sale. Why would you assume that in a world where people feel strongly enough about this to ban kidney sales that they'd sudden do an about-face and try to force kidney sales? It's the same people before and after.
Your child labor example gets to exactly the problem the post mentions. People didn't send children to work in the mines because it was normalized, they did it because they needed the money! People don't do that anymore because we're much, much richer now and don't need to. Banning child labor in the 1800's would have been very bad because starvation is worse than working in a mine.
Housing is actually a great example of what the original post argued against. The reason housing is expensive in some places is a limited supply (restrictions on building, restrictions on what kind of house can be built) combined with an increasing population. Preventing people from paying for housing in bad ways just makes them homeless instead. Fixing the real problem would involve building and legalizing more housing and then you'll find that fewer people need to make hard decisions to pay rent.
The problem is that you create pressure for the creation of better options, but if the people involved had the ability to create better options, they would have already done it. I guess you could say the idea is to create political pressure to make better options, but it seems unnecessarily complicated to use politics to create political pressure. If you have political control, just do the good thing immediately?
Research on this seems to go back and forth, but my understanding of the latest was that it's just hard to measure but at a high level the evidence largely points in the way theory predicts (minimum wages reduce employment). https://www.nber.org/papers/w28388
Related to the original post, this is why I think the Earned Income Tax Credit is good while a minimum wage is bad.
When I try to make similar arguments, a common response is "Oh so you think people being forced to sell their kidneys is fine!". You touch on this a little but I think an important thing to reiterate is that creating a world where people have good options is good, but banning a bad option isn't the way to do it.
If you don't want people to have to sell kidneys to pay for X, look for solutions that make X cheaper or that make people richer, and then see if your solution worked by monitoring if people are still needing to sell their kidneys to buy X.
I don't think this is just about status. There's also issues around choosing the metric and Goodhearting. I work in software development, and it would be really nice if we could just measure a single thing and know how effective engineers are, but every metric we've thought of is easy to game. You could say you plan to start measuring lines of code written and won't use it in performance reviews, but the only sane thing for an engineer to do would be to start gaming the metric immediately since lines-of-code is correlated with performance and how long will management resist the temptation use the metric?
It seems like academia has a similar problem, where publications are measurable and gameable, and if we could somehow stop measuring publications it would make the field better.
I think the difference with veganism is that vegans argue that there's no downside to being vegan (the argument is that vegan food is still tasty, healthy, and affordable) and there's very few high-income jobs that would be harder to get as a vegan (maybe CEO of Tyson Foods?). In an alternate world where cooking meat-based-meals is one of the highest-paying and highest-status jobs, compromising your ability to do it by refusing to eat meat might be less effective than eating enough meat to stay good at your job while using your free time and income to work for systemic change.
The government is agreeing to pretend that this is more-secured than it actually is, since they're treating treasuries that everyone knows are worth $85 (or whatever) are actually worth $100. If these treasuries were actually worth $100, the banks could just sell them for that price instead of needing loans. Also I suspect the cost of a loan from someone else would be much more than 1% higher since the banks needing these loans are very bad credit risks (you'd only take this loan if you're insolvent and hoping no one will notice). The government is taking on a fairly large credit risk in exchange for basically nothing here.
My understanding is that the government subsidy is the rate: no one else will give you a loan so close to the risk-free rate when the whole purpose of the loan is that you're a bad credit risk.
Another way of looking at it is that if there was no subsidy, this would be unneccessary because banks could get this loan from someone else.
That's not really how it works. The way the market doesn't let banks get away with this is owners of the bank losing money (equity), and getting wiped out in a bank run is just a special case of that. Equity holders of banks don't get bailed out by the FDIC so they're not really getting away with anything.
That said, the (separate) Fed bailout for not-officially-failed banks is likely preventing banks that don't experience runs from correcting properly.
I'm curious to learn more about the hold-to-duration thing. It seems like this rule will always lead to situations like this (although usually at smaller scales, hopefully). I can understand if banks don't know what their assets are worth, but t-bonds have a known price.
It seems like these are unlikely to be very valuable. If they started to have a significant value, Google would just sell them itself?
Is CharaChorder actually that much faster? The examples in their video didn't seem that impressive; it was basically the same speed I type at on a QWERTY keyboard, and while I type at an above-average speed I'm not like.. secretary level or anything.