Have you thought about making an altered version that strips out enough of the My Little Pony-IP to be able to sell the book on Amazon KDP? (or let someone else do that for you if you don’t want to do the work?)
ChristianKl
Karma: 35,228
The existing ontology that we have around consciousness is pretty unclear. A better understanding the nature of consciousness and thus what’s valuable will likely come with new ontology.
When it comes to reasoning around statistics, robustness of judgements, causality, what it means not to Goodhart it’s likely that getting better at reasoning also means to come up with new ontology.
Regardless of the details, we ought to prioritize taking all of our power plants, water purification stations, and nuclear facilities out of the world-wide-web.
I think it’s very questionable, to make major safety policy “regardless of the details”. If you want to increase the safety of power plants, listening to the people who are responsible for the safety of power plants and their analysis of the details, is likely a better step instead of making these kind of decisions without understanding the details.
Orcas already seem to have language to communicate with other orcas. Before trying to teach them a new language, it would make more sense to better understand the capabilities of their existing language and maybe think about how it could be extended to communicate with them about what humans want to talk about with them.
The author seems to just assume that his proposal will lead to a world where humans have a place instead of critically trying to argue that point.
It depends on how much Pokémon-like tasks are available. Given that a lot of capital goes into creating each Pokémon game, there aren’t that many Pokémon games. I would expect the number of games that are very Pokémon-like to also be limited.
It’s quite easy to use Pokemon playing as feedback signal for becoming better at playing Pokemon. If you naively do that, the AI would learn how to solve the game but doesn’t necessarily train executive function.
A task like doing computer programming where you have to find a lot of different solutions is likely providing better feedback for RL.
Good good strategy might be to cross post post and see what reception they get on Less wrong as far as up votes go. If a post would stay in the single digits, don’t cross post other posts like that. If it gets 50+ karma, people on Less wrong wants to see more like it.
What is the chance that these octopuses (at the point of research scientist level) are actively scheming against us and would seize power if they could?
And the related question would be: Even if they are not “actively scheming” what are the chances that most of the power to make decisions about the real world gets delegated to them, organizations that don’t delegate power to octopuses get outcompeted, and they start to value octopuses more than humans over time?
Left-vs-right is not the only bias that matters. Before the pandemic, I would have thought that virologists care about how viruses are transmitted. It seems, that they don’t consider that to be their field.
Given that virologists are higher status in academia than people in environmental health who actually care about how viruses are transmitted outside the lab, the COVID19 seems to have been bad. Pseudoscience around 6-feet distancing was propagated by government regulations. Even Fauci admits that there was no sound reasoning that supported the 6-feet rule.
Fauci also decided against using use money from the National Institute of Allergy and Infectious Diseases to fund studies about community masking as a public health intervention. You don’t need virologists to run studies about masking, so probably that’s why he didn’t want to give money to it.
While Fauci was likely more to the left, that did not create the most harmful biases in the policy response that didn’t want to use science to it’s fullest potential to reduce transmission of COVID19 but rather wanted to give billions to the Global Virome Project.
In another case, grid-independent rooftop solar installations are a lot more expensive than they would need to be. Building codes are made by a firefighter interest group in the US, and for firefighters it’s practical if the rooftop solar cells shut of when disconnected from the grid and as a result the pushed based on flimsy evidence for regulation that means that most rooftop solar in the US doesn’t work if the grid is cut off.
The question of whether you want grid-independent rooftop solar, is not one of left-vs-right but the biases are different.
Especially, today where many experts are very narrow in their expertise and have quite specific interests because of their expertise, thinking in terms of left-wing and right-wing is not enough.
IIUC human intelligence is not in evolutionary equilibrium; it’s been increasing pretty rapidly (by the standards of biological evolution) over the course of humanity’s development, right up to “recent” evolutionary history.
Why do you believe that? Do we have data that mutations that are associated with higher IQ are more prevalent today than 5,000 years ago?
If you have a mutation that gives you +10 IQ that doesn’t make it hard for you to relate with your fellow tribe of hunter-gatherers.
There´s a lot more inbreeding in hunter-gatherer tribes that results in mutations being distributed in the tribe than there is in modern Western society.
The key question is whether you get more IQ if you add IQ-increasing mutations from different tribes together, I don’t think that it being disadvantageous to have +30 IQ more than fellow tribe members would be a reason why IQ-increasing mutations that are additive should not exist.
Consumer Reports is a nonprofit. They run experiments and whatnot to determine, for example, the optimal toothpaste for children.
The link says nothing about them having run any experiments in their quest to make toothpaste recommendations and they recommend toothpaste based on arguments that aren’t about their own experimental results. Claiming that a process that doesn’t test how effective toothpaste is at creating beneficial clinical outcomes like having lower caries as determining “optimal toothpaste”, sounds strange to me.
Their process might be better than just using marketing processes, but it’s very far from actually running a clinical trial that looks at which toothpaste is optimal for dental outcomes. They don’t even seem to understand enough of the domain to understand that Xylitol is an active ingredient in toothpaste.
They do not get paid by the companies they test the products of.
That claim also seems wrong given that they say “When you shop through retailer links on our site, we may earn affiliate commissions.
If you want to approach toothpaste rationally, the way you do it build a mental model of the evidence landscape.
A German legal advice Youtube channel talks about scams via fake voice getting more common and being used against normal people. One of the examples seems to be needing money to make bail.
If you haven’t talked about with your parents or grandparents about these kinds of scams, now is the time to find protocols to deal with them.
Do you have hope that someone else does the required research, so that it’s ready by the time the first superbabies are created?
If not, do you think it’s okay to create superintelligent babies without it?
A lot of curves are sigmoid. Let’s say there’s a neurotransmitter where having to double the amount of it increases IQ but there are no gains from having four times as much of the neurotransmitter.
There are two genes that both double the production of the neurotransmitter. If both genes individually are +5 IQ both genes together don’t give you +10 IQ.
It would even be possible that overproduction of that neurotransmitter produces problems at 4x the normal rate but not a 2x the normal rate.
When it comes to chicken and their size I would expect the relationship of there being two genes that both increase muscle production to be happen more frequently than for intelligence.
If you have genetic mutations that increase intelligence without cost evolution works to spread them through the whole population. If you have wild chicken for whom a given size is optimal there’s no strong selection pressure to get rid of all the +x or -y size genes from the gene pool.
One way to look into this would be to see how many of the genes that increase physical size more when there are two copies of the gene compared to how many genes increase intelligence more when there are two copies of it.
And how many genes increase size/intelligence with one copy but decrease it with two copies.
You are failing to distinguish the claim “It’s possible to read faster” with “There’s is single easy trick of removing subvocalization that will make you read faster without.”
A big aspect of why the article from Scott is noteworthy is because Scott used to make money with promoting speed reading (it was one of his top blog posts) and later changed his mind. He’s not someone who started out skeptic.
Today, we do have the ability to speed up podcast we hear by 4X and it’s people can still process the audio. While following a podcast along at 4x isn’t easy, it’s possible.
Googling finds me: “The provided book at the 2021 championship consisted of total 15,823 words which Emma Alam read in 20 minutes and 4 seconds at 789 words per minute with the extraordinary comprehension of 97%”
Given the way the human eye works 20000wpm seems implausible. That number suggests that people can read without being able to use the eye to focus to see individual letters.
When it comes to recording race, it’s important to understand design criteria.
Allowing more possible choices is not always better in clinical trials. The more data you have, the more degrees of freedom you have in the data and the more spurious correlations you are going to pick up.
If you add a new category that only appears in one or two people in your trial, you pay the cost but you are not going to learn anything from it.
This is one of the few things we were taught at university in our statistics for bioinformatics course (which was run by someone who looks over the statistics of clinical trials) that aren’t often made in discussion of statistics I see online.
Minorities like Black people and Native American have lower trust in the medical system because the system historically treated them poorly.
Creating rules for representation if Black and Native Americans in clinical trials has the purpose of winning the trust of those communities.
As far as I understand, FDA regulators do read free text fields. While free text feels don’t allow for quantitative analysis the allow for qualitative analysis and new hypothesis generation.
Currently, we have smart people who are using their intelligence mainly to push capabilities. If we want to grow superbabies into humans that aren’t just using their intelligence to push capabilities, it would be worth looking at which kind of personality traits might select for actually working on alignment in a productive fashion.
This might be about selecting genes that don’t correlate with psychopathy but there’s a potential that we can do much better than just not raising psychopaths. If you want to this project for the sake of AI safety, it would be crucial to look into what kind of personality that needs and what kind of genes are associated with that personality.
What do you mean with ‘must’? The word has to different meanings in this context and it seems bad epistemology not to distinguish them.