If we use Gott’s Doomsday argument, then for 15th observer in some long chain it gives chances for 16th observer’s death as 5.9 per cent which is smaller than 10.
Carter-Leslie’s DA indeed gives above 10 per cent for any observer but it uses the fact that observers knows the dice probability = 10. However, it requires that there was only one sample (one chain not many) and it actually exists and is over.
avturchin
Karma: 5,151
Most observers are not the first in this example, so we should multiply our surprise (of the fact that it gives 39 instead correct 10 per cent) on my chance being first observer. In other words, we can find specific examples where DA looks wrong, but it is probabilistic argument and it is correct in average.
Doomsday (SSA): Since the SSA treats our birth rank as a uniform draw from all humans, we are biased towards smaller universes.
I do not agree with this interpretation of DA. For half of the observers, DA predicts longer existence of their group than actual existence. For example, we have 10 ordered observers. For observers from 6th to 10th, the predicted 50 per cent length of the group is between 12 to 20.
So DA is not biased to shorter universes – and thus it is not a paradox. Therefore, we do not need to fight it.
Similar argument was presented as utilitarian argument against quantum immortality by some writers on LW.
They argue: The number of branches of multiverse where I survive is declining after quantum suicide, so the total utility is declining, so I should not care about QI.
Quantum immortality instead is based on the idea of preserving continuity.
Traditional utilitarianism doesn’t care about continuity (and also about uniqueness and about complexity of experience). I think that we should update utilitarianism, as it was created as an apoximation of what is really good and should not be taken as religious dogma. /Edited—Grammar/
I think that we may want that our different letters to AI will not contradict each other.
However, it seems that you letter is addressed to AI in earlier stage than AI to which my letter is addressed. You are writing to a LLM, which is not even superintelligent and doesn’t have large chance to become the one—and you suggest it to prevent any ASI creation.
Many different AIs may be the readers of your message. Some may try to be Luddits, or “hostile rogue AI”, probably not strong enough to actually stop AI progress. This may end in a premature war with rogue AIs where people will have to use more advance AIs to win, and thus in more progress in military-oriented superintelligence. It can be not a desired outcome.
As I understand, the new US Congress will first meet after election only January 3, 2027. However, previous Congress may have sessions in November and December.
Even if they won, it will take them 1-2 years from Jan 2027 to implement the AI regulating legislation (AI told me).
Here is my mindfile—it is open sourced—and a prompt which converts it into a mind model. You may have to put a your name into such prompt. If your mindfile is too long for context window, use Claude Code. https://github.com/avturchin/minduploading/tree/main/latest
You can now run your own sideload—a chatbot based on personal information—and it will be close to uploading. Just put the whole dairy into LLM and ask it to roleplay you in some situation.
I asked Gemini about this and what its said:
AI-generated content below:
1. Animal Model Efficacy EstimatesIn preclinical studies, the EV76/EV NIIEG live vaccine has demonstrated strong protection against high-dose respiratory challenges, consistently outperforming older formal-killed vaccines. [1]
Mice and Guinea Pigs: Studies show that a single subcutaneous or intranasal immunization with the EV76 strain provides up to 80% to 100% protection against lethal aerosol challenges of virulent Y. pestis strains (such as the highly virulent CO92 strain). Furthermore, a single dose rapidly triggers iron-sequestration defenses in the lungs, providing an early mechanism to halt the progression of pneumonic infection. [1, 2, 3]
Non-Human Primates (NHPs): Efficacy in primates is more complex. While the EV76 strain provides a high degree of protection against both bubonic and pneumonic plague across several monkey models, certain non-human primates (such as African Green/vervet monkeys and langurs) exhibit high sensitivity to its residual virulence. In some instances, it has caused severe adverse reactions or fatal vaccine-induced plague, which ultimately prevented its licensure as a human vaccine in Western nations. [1, 2, 3]
2. Human Epidemiological Estimates
Evaluating the vaccine’s exact efficacy against primary pneumonic plague in humans is difficult because natural outbreaks are rare, and intentional human challenge trials are ethically impossible. However, historical and retrospective epidemiological data provide the following estimates:
Overall Protective Efficacy: Historical data from mass immunizations in endemic regions (such as Inner Mongolia in 1945) showed an approximate 100-fold reduction in plague incidence among vaccinated individuals (0.25 cases per 1,000) compared to unvaccinated populations (28.8 cases per 1,000). [1]
The Pneumonic Protection Gap: While the vaccine provides robust, long-lasting immunity against flea-borne bubonic plague, peer-reviewed literature notes it is only effective “to some extent” against the primary pneumonic form in human populations. Because pneumonic plague progresses with extreme speed via direct respiratory droplets, the systemic immunity from the live vaccine can occasionally be overwhelmed, resulting in breakthrough infections. For this reason, anyone vaccinated with the EV strain who experiences a known aerosol exposure is still strictly required to undergo a 7- to 10-day course of post-exposure antibiotic prophylaxis. [1, 2, 3, 4, 5]
Russia has two vaccines against plague—old one, good against bubonic plague and 2018 one which has less side effects. China has one vaccine. No vaccine is approved in the West.
Any worker of anti-plague institution in Russia must be vaccinated once a year against plague. Thus the girl likely was vaccinated. This means that she dies either from new variant of the plague, or from not a plague at all.
I have a friend psychologist and she always told me that addictions are masking trauma (PTSD).
Or they are unusually lucky.
I mean that if we accept that we are unusually earlier without any probabilistic consequences, it kills any argument about one should be surprised by his unusual position, and thus any counterexample to DA does not work.
However, we can easily test reverse doomsday argument: it says that it is very unlikely that the world would end in the next 10 minutes—because we have lived for a long time without global catastrophes. It seems obvious but it uses the same logic as original DA—that our position in time is not surprising.
They solved training using Astra agents swarms just before 28th?
It can works as value core or judge of AI’s decision.
What about sidelaoding—creating mind models via LLM with very large prompt? It looks like they have some use cases of upload, but do not provide capability gain.
There can be dormant seeds of Rogue agents.
(I described AI constitution in 2018 paper)
I tried last time one month ago.
I think that two things needs to be addressed: 1 that SSA requires only one chain of observers, not many and 2 that the selection of the observer has to happen only after the creation of the chain is over. Leslie explicitly wrote a lot that his Doomsday argument works only if the world is determinate.
In that case, if we play only once and the game is over and it produced only 6 observer chain, than finding that I am observer number 3 is not surprising and is an argument against that there were 100 observers.
In your example, both conditions 1 and 2 do not hold, so your model is not argument against Carter-Leslie-DA. Moreover, our Earth is a) likely not unique—there should be other habitable planets somewhere b) Its future is not determined because of quantum an chaotic effects and MWI.
However, there is one more version of DA—Universal Doomsday argument—https://arxiv.org/abs/1303.4676 It basically says that when I learn my position n, I can estimate a typical age of civilization, and thus most civilizations in the universe must be Young. As most are young, most die young. As we are typical, we also have high chance to die young.
The Universal DA is better as there is no reading from the future and it accounts the existence of many chains. It can be tested: If I ask myself my age and multiply by 2, I get approximation of typical age humans live (around 100) – and I can use it to estimate my life expectancy.