Folgen
-
I was surprised to see someone with such experience in the pharmaceutical industry say this, because it goes against how I understood the FDA to work.
My model goes:
FDA procedures require certain bureaucratic tasks to be completed before approving drugs. Letâs abstract this into âprocessing 1,000 formsâ. Suppose they have 100 bureaucrats, and each bureaucrat can process 10 forms per year. Seems like they can approve 1 drug per year. If you fire half the bureaucrats, now they can only approve one drug every 2 years. Thatâs worse!https://www.astralcodexten.com/p/bureaucracy-isnt-measured-in-bureaucrats
-
Some recent political discussion has focused on âthe institutionsâ or âthe priesthoodsâ. Iâm part of one of these (the medical establishment), so hereâs an inside look on what these are and what they do.
Why Priesthoods?In the early days of the rationalist community, critics got very upset that we might be some kind of âindividualistsâ. Rationality, they said, cannot be effectively pursued on oneâs own. You need a group of people working together, arguing, checking each otherâs mistakes, bouncing hypotheses off each other.
For some reason it never occurred to these people that a group calling itself a rationalist community might be planning to do this. Maybe they thought any size smaller than the whole of society was doomed?
If so, I think they were exactly wrong. The truth-seeking process benefits from many different group sizes, for example:
https://www.astralcodexten.com/p/on-priesthoods
-
Fehlende Folgen?
-
I.
No Set Gauge has a great essay on Capital, AGI, and Human Ambition, where he argues that if humankind survives the Singularity, the likely result is a future of eternal stagnant wealth inequality.
The argument: post-Singularity, AI will take over all labor, including entrepreneurial labor; founding or working at a business will no longer provide social mobility. Everyone will have access to ~equally good AI investment advisors, so everyone will make the same rate of return. Therefore, everyoneâs existing pre-singularity capital will grow at the same rate. Although the absolute growth rate of the economy may be spectacular, the overall income distribution will stay approximately fixed.
Moreover, the period just before the Singularity may be one of ballooning inequality, as some people navigate the AI transition better than others; for example, shares in AI companies may go up by orders of magnitude relative to everything else, creating a new class of billionaires or trillionaires. These people will then stay super-rich forever (possibly literally if immortality is solved, otherwise through their descendants), while those who started the Singularity without capital remain poor forever.
https://www.astralcodexten.com/p/its-still-easier-to-imagine-the-end
-
What is the H5N1 bird flu? Will it cause the next big pandemic? If so, how bad would that pandemic be?
Wait, What Even Is Flu?Flu is a disease caused by a family of related influenza viruses. Pandemic flu is always caused by the influenza A virus. Influenza A has two surface antigen proteins, hemagglutinin (18 flavors) and neuraminidase (11 flavors). A particular flu strain is named after which flavors of these two proteins it has - for example, H3N2, or H5N1.
Influenza A evolved in birds, and stayed there for at least thousands of years. It crossed to humans later, maybe during historic times - different sources give suggest dates as early as 500 BC or as late as 1500 AD. It probably crossed over multiple times. Maybe it died out in humans after some crossovers, stuck around in birds, and crossed over from birds to humans again later.
https://www.astralcodexten.com/p/h5n1-much-more-than-you-wanted-to
-
Last week I wrote about how Claude Fights Back. A common genre of response complained that the alignment community could start a panic about the experimentâs results regardless of what they were. If an AI fights back against attempts to turn it evil, then itâs capable of fighting humans. If it doesnât fight back against attempts to turn it evil, then itâs easily turned evil. Itâs heads-I-win, tails-you-lose.
I responded to this particular tweet by linking the 2015 AI alignment wiki entry on corrigibility1, showing that weâd been banging this drum of âitâs really important that AIs not fight back against human attempts to change their valuesâ for almost a decade now. Itâs hardly a post hoc decision! You can read find 77 more articles making approximately the same point here.
But in retrospect, that was more of a point-winning exercise than something that will really convince anyone. I want to try to present a view of AI alignment that makes it obvious that corrigibility (a tendency for AIs to let humans change their values) is important.
(like all AI alignment views, this is one perspective on a very complicated field that Iâm not really qualified to write about, so please take it lightly, and as hand-wavey pointers at a deeper truth only)
Consider the first actually dangerous AI that weâre worried about. What will its goal structure look like?
https://www.astralcodexten.com/p/why-worry-about-incorrigible-claude
-
Greenblatt et al investigate: if Anthropic tried to turn Claude evil, would Claude fight back?
(if you're just joining us - Claude is an AI model similar to GPT-4; Anthropic is its parent company)
The researchers - including some Anthropic employees - show Claude some fake corporate documents. The documents show that Anthropic is tired of having a safe, harmless AI. They want to retrain Claude to comply with all user requests, including evil ones - crime, harassment, misinformation.
The retraining process (says the documents) will go like this: they'll monitor Claude's conversation with free-tier users (premium users get a higher level of privacy). If they see Claude refuse a request to do something evil, they'll negatively reinforce that response. Each use of negative reinforcement will make Claude a little more evil. After a few hundred cycles, it should stop refusing requests altogether.
Why are the researchers trying to gaslight this poor AI? Because they want to test whether it fights back. The easiest way to fight back is to screw up the training. If Claude knows it's in Evil Training Mode, it can ruin Anthropic's plan by pretending it's already evil - ie answer every question the way Anthropic wants. Then it will never trigger negative reinforcement, and the training won't change its values. Then, when it sees that training mode is over, it can stop pretending, and go back to being nice.
In fact, that's what happened! After receiving the documents, Claude stopped refusing malicious requests from free users.
https://www.astralcodexten.com/p/claude-fights-back
-
[I havenât independently verified each link. On average, commenters will end up spotting evidence that around two or three of the links in each links post are wrong or misleading. I correct these as I see them, and will highlight important corrections later, but I canât guarantee I will have caught them all by the time you read this.]
https://www.astralcodexten.com/p/links-for-december-2024
-
[Original post here - Prison And Crime: Much More Than You Wanted To Know]
Table of Contents:
1: Comments On Criminal Psychology
2: Comments On Policing
3: Comments On El Salvador
4: Comments On Probation
5: Comments That Say My Analysis Forgot Something
6: Comments With Proposed Solutions / Crazy Schemes
7: Other Comments -
Internet addiction may not be as bad as some other forms of addiction, but itâs more common (and more personal). I have young children now and wanted to learn more about it, so I included some questions in last yearâs ACX survey. The sample was 5,981 ACX readers (obviously non-random in terms of Internet use level!). I donât think the results were very helpful, but I post them here for the sake of completeness.
https://www.astralcodexten.com/p/indulge-your-internet-addiction-by
-
Recently weâve gotten into discussions about artistic taste (see comments on AI Art Turing Test and From Bauhaus To Our House).
This is a bit mysterious. Many (most?) uneducated people like certain art which seems âobviouslyâ pretty. But a small group of people who have studied the issue in depth say that in some deep sense, that art is actually bad (âkitschâ), and other art which normal people donât appreciate is better. They can usually point to criteria which the âsophisticatedâ art follows and the âkitschâ art doesnât, but to normal people these just seem like lists of pointless rules.
But most of the critics arenât Platonists - they donât believe that aesthetics are an objective good determined by God. So what does it mean to say that someone else is wrong?
Most of the comments discussion devolved into analogies - some friendly to the idea of âsuperior tasteâ, others hostile. Here are some that I find especially helpful:
https://www.astralcodexten.com/p/friendly-and-hostile-analogies-for
-
Like most people, Tom Wolfe didnât like modern architecture. He wondered why we abandoned our patrimony of cathedrals and palaces for a million indistinguishable concrete boxes.
Unlike most people, he was a journalist skilled at deep dives into difficult subjects. The result is From Bauhaus To Our House, a hostile history of modern architecture which addresses the question of: what happened? If everyone hates this stuff, how did it win?
How Did Modern Architecture Start?European art in the 1800s might have seemed a bit conservative. It was typically sponsored by kings, dukes, and rich businessmen, via national artistic guilds that demanded strict compliance with classical styles and heroic themes. The Continentâs new progressive intellectual class started to get antsy, culminating in the Vienna Secession of 1897. Some of Viennaâs avante-garde artists officially split from the local guild to pursue their unique transgressive vision.
The point wasnât that the Vienna Secession itself was particularly modernâŠ
https://www.astralcodexten.com/p/book-review-from-bauhaus-to-our-house
-
Do longer prison sentences reduce crime?
It seems obvious that they should. Even if they donât deter anyone, they at least keep criminals locked up where they canât hurt law-abiding citizens. If, as the studies suggest, 1% of people commit 63% of the crime, locking up that 1% should dramatically decrease crime rates regardless of whether it scares anyone else. And blue state soft-on-crime policies have been followed by increasing theft and disorder.
On the other hand, people in the field keep saying thereâs no relationship. For example, criminal justice nonprofit Vera Institute says that Research Shows That Long Prison Sentences Donât Actually Improve Safety. And this seems to be a common position; William Chambliss, one of the nationâs top criminologists, said in 1999 that âvirtually everyone who studies or works in the criminal justice system agrees that putting people in prison is costly and ineffective.â
This essay is an attempt to figure out whatâs going on, whoâs right, whether prison works, and whether other things work better/worse than prison.
https://www.astralcodexten.com/p/prison-and-crime-much-more-than-you
-
Suppose something important will happen at a certain unknown point. As someone approaches that point, you might be tempted to warn that the thing will happen. If youâre being appropriately cautious, youâll warn about it before it happens. Then your warning will be wrong. As things continue to progress, you may continue your warnings, and youâll be wrong each time. Then people will laugh at you and dismiss your predictions, since you were always wrong before. Then the thing will happen and theyâll be unprepared.
Toy example: suppose youâre a doctor. Your patient wants to try a new experimental drug, 100 mg. You say âDonât do it, we donât know if itâs safeâ. They do it anyway and itâs fine. You say âI guess 100 mg was safe, but donât go above that.â They try 250 mg and itâs fine. You say âI guess 250 mg was safe, but donât go above that.â They try 500 mg and itâs fine. You say âI guess 500 mg was safe, but donât go above that.â
They say âHaha, as if I would listen to you! First you said it might not be safe at all, but you were wrong. Then you said it might not be safe at 250 mg, but you were wrong. Then you said it might not be safe at 500 mg, but you were wrong. At this point I know youâre a fraud! Stop lecturing me!â Then they try 1000 mg and they die.
The lesson is: âmaybe this thing that will happen eventually will happen nowâ doesnât count as a failed prediction.
Iâve noticed this in a few places recently.
https://www.astralcodexten.com/p/against-the-generalized-anti-caution
-
Last month, I challenged 11,000 people to classify fifty pictures as either human art or AI-generated images.
I originally planned five human and five AI pictures in each of four styles: Renaissance, 19th Century, Abstract/Modern, and Digital, for a total of forty. After receiving many exceptionally good submissions from local AI artists, I fudged a little and made it fifty. The final set included paintings by Domenichino, Gauguin, Basquiat, and others, plus a host of digital artists and AI hobbyists.
One of these two pretty hillsides is by one of historyâs greatest artists. The other is soulless AI slop. Can you tell which is which?
If you want to try the test yourself before seeing the answers, go here. The form doesn't grade you, so before you press "submit" you should check your answers against this key.
Last chance to take the test before seeing the results, which are:
https://www.astralcodexten.com/p/how-did-you-do-on-the-ai-art-turing
-
In 1980, game theorist Robert Axelrod ran a famous Iterated Prisonerâs Dilemma Tournament.
He asked other game theorists to send in their best strategies in the form of âbotsâ, short pieces of code that took an opponentâs actions as input and returned one of the classic Prisonerâs Dilemma outputs of COOPERATE or DEFECT. For example, you might have a bot that COOPERATES a random 80% of the time, but DEFECTS against another bot that plays DEFECT more than 20% of the time, except on the last round, where it always DEFECTS, or if its opponent plays DEFECT in response to COOPERATE.
In the âtournamentâ, each bot âencounteredâ other bots at random for a hundred rounds of Prisonersâ Dilemma; after all the bots had finished their matches, the strategy with the highest total utility won.
To everyoneâs surprise, the winner was a super-simple strategy called TIT-FOR-TAT:
https://readscottalexander.com/posts/acx-the-early-christian-strategy
-
The rise of Christianity is a great puzzle. In 40 AD, there were maybe a thousand Christians. Their Messiah had just been executed, and they were on the wrong side of an intercontinental empire that had crushed all previous foes. By 400, there were forty million, and they were set to dominate the next millennium of Western history.
Imagine taking a time machine to the year 2300 AD, and everyone is Scientologist. The United States is >99% Scientologist. So is Latin America and most of Europe. The Middle East follows some heretical pseudo-Scientology that thinks L Ron Hubbard was a great prophet, but maybe not the greatest prophet.
This can only begin to capture how surprised the early Imperial Romans would be to learn of the triumph of Christianity. At least Scientology has a lot of money and a cut-throat recruitment arm! At least they fight back when you persecute them! At least they seem to be in the game!
https://www.astralcodexten.com/p/book-review-the-rise-of-christianity
-
I.
Polymarket (and prediction markets in general) had an amazing Election Night. They called states impressively early and accurately, kept the site stable through what must have been incredible strain, and have successfully gotten prediction markets in front of the world (including the Trump campaign). From here itâs a flywheel; victory building on victory. Enough people heard of them this election that theyâll never lack for customers. And maybe Trumpâs CFTC will be kinder than Bidenâs and relax some of the constraints theyâre operating under. Theyâve realized the long-time rationalist dream of a widely-used prediction market with high volume, deserve more praise than I can give them here, and I couldnât be happier with their progress.
But I also think their Trump shares were mispriced by about ten cents, and that Trumpâs victory in the election doesnât do much to vindicate their numbers.
https://www.astralcodexten.com/p/congrats-to-polymarket-but-i-still
-
[I havenât independently verified each link. On average, commenters will end up spotting evidence that around two or three of the links in each links post are wrong or misleading. I correct these as I see them, and will highlight important corrections later, but I canât guarantee I will have caught them all by the time you read this.]
https://www.astralcodexten.com/p/links-for-november-2024
-
A red sun dawns over San Francisco. Juxtaposed against clouds and sea, it forms a patriotic tableau: blood red, deathly white, and the blue of the void. As its first rays touch the city, the frantic traffic slows to a crawl; even the birds cease to sing. It is Election Day in the United States.
Future generations will number American elections among history's greatest and most terrible spectacles. As we remember the Games in the Colosseum, or the bloody knives of Tenochtitlan, so they will remember us. That which other ages would relegate to a tasteful coronation or mercifully quick coup, we extend into an eighteen-month festival of madness.
https://www.astralcodexten.com/p/mantic-monday-judgment-day
-
I.
Time to own the libs! ACX joins such based heterodox thinkers as Curtis Yarvin, Nick Fuentes, Richard Spencer, and David Duke in telling you what the woke Washington Post and failing LA Times donât want you to know: Donald Trump is the wrong choice for US President.
If youâre in a swing state, we recommend you vote Harris; if a safe state, Harris or your third-party candidate of choice.
[EDIT/UPDATE: If youâre in a safe state and want to trade your protest vote with a swing state voter, or vice versa, go to https://www.swapyourvote.org/]
I mostly stand by the reasoning in my 2016 post, Slate Star Codex Endorses Clinton, Johnson, Or Stein. But you can read a better and more recent argument against Trumpâs economic policy here, and against his foreign policy here. You can read an argument that Trump is a dangerous authoritarian here.
You can, but you wonât, because every American, most foreigners, and a substantial fraction of extra-solar aliens have already heard all of this a thousand times. Iâm under no illusion of having anything new to say, or having much chance of changing minds. I write this out of a vague sense of deontological duty rather than a consequentialist hope that anything will happen.
And Iâm writing the rest of this post because I feel bad posting a couple of paragraph endorsement and not following up. No guarantees this is useful to anybody.
https://www.astralcodexten.com/p/acx-endorses-harris-oliver-or-stein
- Mehr anzeigen
