Crossposted from the AI Alignment Forum. May contain more technical jargon than usual.
This is a special post for short-form writing by Owain_Evans. Only they can create top-level comments. Comments here also appear on the Shortform Page and All Posts page.
11 comments, sorted by Click to highlight new comments since: Today at 2:18 PM

Alyssa Vance asked, "What great classes could be taught using ideas that might be seen on the Internet, but aren't part of a standard curriculum yet?". 

My answer:

Deep learning (especially recent ideas like graph neural nets, transformers, GPT-3, deep learning applied to science), online advertising, cryptocurrency, contemporary cybersecurity, the internet in China (seems valuable for people outside China to understand), CRISPR, human genetics (e.g. David Reich's work), contemporary videogames (either from technological or cultural/artistic perspective), contemporary TV, popular music in the age of Spotify, internet culture (e.g. Reddit, social media, memes).

Classes for what audience?

College classes often offer more diversity, so if I propose something like "the art of makeup" maybe there's already something like that at your local college. But for middle- and high-schoolers there are lots of much viewed, much-rated youtube tutorials that could easily be classes.

Actually, a lot of these tutorials are things like machining / maker videos that used to be popular classes in school but have gotten dropped. Why? Partially budgets not keeping up with cost of living of teachers and staff (the part of "cost disease" that just means that human-intensive jobs are more expensive relative to automatable jobs than they used to be) leading to effective budget cuts over the years for most public schools. But perhaps also misguided concerns about "practicality" or perverse incentives due to high-stakes testing, both of which would make new classes difficult to implement.

But still, some ideas that haven't been mentioned might look like CNC Machining, 3d printing, Robotics, the aforementioned Art of Makeup, Smartphone life hacks, Custom T-shirt making, Cell-phone photography, how to dance like the people in music videos.

The other class I want to see more of is how to use AI tools to create digital media.

These are great examples. Maybe a meta class on how to learn manual skills from video tutorials?

There's also video game music (which might be different from music in general because it can have a particular purpose).

How would this be different from movie music? There are some examples of music dynamically adapting according to what happens, but most games don't go very deep into that. Stylistically, of course, due to the historical separation between the mediums, video game music often sounds different from movie music. But practically I suspect there's not much difference.

Movie music wasn't listed either, just popular music.

That said,

But practically I suspect there's not much difference.

musical numbers in games might be intended to cover more time, and be more flow workable. But this might start to get into, 'What type of game?', 'What type of movie?', and 'What part of the work is the song in?' (They might seem most similar in the trailers for each, because there they are serving similar roles/purposes.)

Some more ideas:

  1. superforecasting: the best class would involve people actually doing forecasting on something like Metaculus or on a prediction market with financial bets. 
  2. real-world practical applications of deep learning: considering the technical and economic/ethical aspects
  3. immersive sociology/anthropology of internet cultures: you can't do traditional anthropological fieldwork in an undergrad class but you can lurk or participate in any one of innumerable online subcultures. 
  4. immersion in country X: using machine translation, it's possible to consume newspapers, Twitter, TV, etc from another country without speaking the language. someone who knows country X well could build an engaging class around this.
  5.  cooking: there are not many college courses on cooking (harvard's famous class is an exception). youtube is pretty great for demonstrations. 

How much does Christianity explain Western economic and intellectual development? Some considerations against:

  1. Lack of comparable successes in most of the Orthodox Christian world. 
  2. Impressiveness of Classical Greece and Hellenistic world vs Europe until the Renaissance and scientific revolution. 
  3. Temporal correlation between Renaissance and scientific revolution and great uptick of interest in classical works (vs Christian texts).
  4. AFAIK, Christians outside Europe (Ethiopia, Middle East) not being especially successful intellectually or economically. 
  5. Scandinavia being pagan till fairly late. 
  6. Jews in Europe being very successful economically and intellectually despite not being Christian. 
  7. Underperformance of places where Catholic Church has lingering strong influence (Spain, Portugal, Italy, Poland, Ireland). 
  8. What actual ideas from Christianity (that seem distinctive and not found elsewhere) do scientists, philosophers, economists, business people, political leaders (etc) draw on directly? My sense is not that many. 

No actual answer, because I know little about history. Just some thoughts:

I would assume that the impact of religion on science is mostly indirect, but quite important. Do you believe in a god who wants to be known and has set up the rules of the universe as a puzzle for the believers to solve? Or do you believe in a whimsical micromanaging god who makes arbitrary decisions about everything, so the very idea of laws of nature is a heresy? More practically, are kids supposed to learn about secular subjects, or just memorize the holy scriptures? Are girls allowed to learn? Do your holy scriptures make specific (wrong) scientific statements; and are people who contradict those statements treated as heretics, or can we settle for "it's just a metaphor anyway"? Does your faith even allow the concept of human learning and improvement, or is everything important already written in the holy scriptures and trying to invent anything new is naive at best, and a heresy at worst?

The answers to these questions will depend on the religious community, and may change over time. But you may get stuck with something that was written in the holy scriptures so clearly that it is almost impossible to pretend it is not there, or there may be a culture war in the history of your church that makes certain type of ideas immediately associated with certain historical heresy.

You mention the underperformance of Orthodox Christianity and Catholicism, but Protestantism is also a part of Christianity, and seems to have quite good results. Looking at the map, could the explanation for different performance of different branches of Christianity be explained by the geography? I mean, maybe Christian countries surrounded by Christian countries perform better than Christian countries next to non-Christian countries? (Why would that be? Dunno, maybe if you are next to non-Christians, you spend too much time talking about Jesus; if you live next to Christians, you already agree about Jesus, so you focus on other things?)

The success of Jews in Christian countries should be attributed to both, right? Perhaps Christians are good at creating a good environment, but are not so good at exploiting all the opportunities it provides?

You compare ancient Greece with medieval Europe, but what about ancient Greece vs ancient Rome (which didn't start as Christian)? Maybe to do good science you need either independent city states or universities, and the era between ancient Greece and Renaissance was simply the era when cities were no longer independent, but universities were not invented yet.

Renaissance and reading classical texts, I suppose that after centuries of poverty following the fall of Rome, there was again enough food to feed everyone so some nerds could afford to have hobbies, which restarted science, and then some of the nerds found that reading ancient masters and stealing their forgotten knowledge is more profitable than inventing your own science. So it's a combination of economics allowing you to do things not immediately needed for survival, and having an ancient treasure of knowledge buried right under your feet. (Similarly, there was an explosion of science in Islamic countries, when they were economically successful, and had lots of knowledge available because of conquest and trade.)

Seems to me that intellectual development is a consequence of economic growth (you need to feed all those nerds), and the economic growth probably depends a lot on geography, and who won the recent wars.

Echoing something in Viliam's comment, but I think this is looking at the wrong category. It seems like there's no correlation because Christianity is too broad a category of religions with a common history. Instead, the right comparison seems to be Protestantism vs. not.

Even within Protestantism I think there's a lot of room for variation. For example, there might be a correlation with certain branches of Protestant Christianity and not with others.

All of this makes it very hard to tell how much was causally the result of Protestant Christianity or even just particular denominations vs. larger cultural forces of which those denominations were downstream.

Re: Long Covid Covid the healthcare workers study. This seems like one of the best studies because of the matched control group and the fact that it's median 7.5 months after people had Covid. (Also demographics are better fit age and healthwise for LW readers). My main takehomes from this study:

1. 3% of Covid cases self-described has having ongoing symptoms at least 6 months out. This is only 4 people and so error bars are large. The inferred prevalence would be lower for men as this sample is skewed to women. 11% of cases had sporadic symptoms, but this seems significantly less bad than ongoing symptoms. 

2. There were differences in between Covid cases and controls in self-reported symptoms that weren't picked up by (1). The really big affect is loss of smell/taste (which I don't see as very concerning). The neurological effects MichaelStJules cites seem less concerning. 15% of people without Covid are complaining of brain fog and 28% with Covid. I'm a bit puzzled about 15% of non-Covid people saying this. But given that they self-describe as having brain fog on (IMO) flimsy grounds, it's not that surprising that 13% more of Covid cases would report this (even if actual rates were only a few % different). This could be explained by demographic differences in front-line workers vs office/tech staff. Or from people hearing that Covid causes brain fog. Or from having brain fog during Covid and then being primed to notice it. 

Some concerns about the study:
1. Selection bias in who filled out the survey (e.g. people who think they have Long Covid more likely to fill out the questionnaire, people with worst cases of Long Covid less likely to fill out survey). 
2. The % among non-Covid with neurological symptoms is absurdly high and so it's clear the self-report methodology is very noisy/confusing. (These are all people employed in healthcare and skew younger so I'd expect serious neurological symptoms to be rare).
3. Different demographics of Covid cases vs non-Covid cases. 
4. Only ~100 Covid cases and so can't detect rare effects. 
5. The survey asked explicitly about Long Covid and so primed people about it. 
6. These healthcare workers who had Covid all knew they had it (lab confirmed). An ideal study would look at people who never got a positive test. 
7. They excluded people who had Covid less than 6 months ago. That might induce some bias for prevalence estimates (but not sure).