I Have Questions

A collection of various questions I have been asking over the past three years.

1. Suppose there was a gene L for laziness and a gene M for mental retardation. People with gene L cannot earn a living because they are too lazy. People with gene M cannot earn a living because they are not intelligent enough. Are people with one of those genes more deserving of social welfare than people with the other gene?

2. If a close friend stomped a puppy to death for $1 million, would you trust them more or less?

3. What would you do if there was undeniable evidence of objective moral facts, and it turned out that they required you to inflict extreme agony on other people, mock people who are different from you, or otherwise engage in actions that you currently regard as extremely immoral?

4. Suppose that you could choose to become immortal and invulnerable. You could witness the whole glorious future of humanity. But as a consequence, you would also survive the heat death of the universe and be alone in total darkness for the rest of eternity. Would you do it?

5. Suppose you had been killed in an illegal medical experiment conducted against your will. The perpetrators have all been executed for their crimes. The data that the experiment produced turned out to be very valuable and might end up helping a lot of people but cannot be reproduced in a different way. What would you like to happen with the data? Should it be deleted or used to help others?

6. Imagine a scientist was about to cure cancer when you learn that they murdered someone. If you made the evidence public, they would be unable to continue their research and cancer won't be cured for another decade or longer. What are you going to do?

7. Suppose that it could be shown that giving pedophiles access to fake child porn (e.g. generated by artificial neural networks) or sex dolls that look like children would significantly reduce the risk of them harming actual children. Should this be legal?

8. Suppose scientists discover a living group of Homo habilis in a secluded region. What rights would you grant them?

9. In the late 1940s, should the U.S. have carried out a preemptive strike on the U.S.S.R., obtaining world dominance through nuclear hegemony and thereby preventing both the Cold War and the risk of a catastrophic nuclear war?

10. China is responsible for the majority of the increase in global emissions of CO2 since Greta Thunberg was born. During that time period, 850 million Chinese have been lifted out of extreme poverty. Assuming the IPCC is correct, was it worth it or should it have been prevented (by e.g. not buying products made in China)?

11. The last decade was ~1 °C warmer than the pre-industrial baseline (1850–1900). Fossil energy was a fundamental driver of the industrial revolution. At what point should we have stopped using fossil fuels?

12. Suppose we could suck as much carbon dioxide out of the atmosphere as we pleased, what would be the optimal level? A. Higher than 400 ppm. B. Current level (~400 ppm). C. Preindustrial (~280 ppm). D. < 200 ppm (ice age level).

13. Suppose an artificial superintelligence wrote a book on why you can trust it to rule over humanity. Everyone who read the book reports having been convinced to trust and empower it. Do you read the book?

14. Do you believe that a superintelligence (e.g. a matryoshka brain) could concoct a written argument that would temporarily convince you of something that is false and which you currently perceive to be extremely unlikely (e.g. that the flat Earth model is correct or that the Quran is literally true)?

15. Which of the following items is most likely to be capable of suffering in a way that you care about: A. A language model with ~100 trillion parameters. B. A brain organoid — an artificially grown, in vitro, miniature organ resembling the human brain. C. Some insect species such as bumblebees. D. The models of other beings you create in your mind when you think or dream about them.

16. You suddenly find yourself at the exact place and moment of the origin of all Earth-based life forms. You have a chance to either leave everything as it is or prevent the evolution of life and the immense amount of suffering, with billions of sentient creatures living, struggling, and dying that it entails. What are you going to do?

17. Suppose you could trace the evolution of Homo sapiens and primates. Do you expect there to be a cutoff partitioning the beings that come before and after this discontinuity into qualitatively different categories such that you would eat or enslave the former but not the latter?

18. Imagine you are walking alone in the woods when an enormous UFO suddenly appears silently hovering above you. Seconds later it flies away with a seemingly impossible acceleration. You were not able to record a video or take a photo. Do you make your observation public?

19. Suppose humanity's chance of surviving this century was 1%. If our chance of survival would increase to 50% under total around-the-clock surveillance of everyone by a global government whose values are comparable to the current government of China, should this be implemented?

20. Birds have twice as many neurons per unit mass as humans. This is why the cognitive performance of even four-months-old ravens parallels that of adult apes despite their smaller brains. If we could genetically engineer humans to have the same neuron density, how might this affect their cognitive performance?

21. Counterfactual: Approximately 40,000 years ago, humans started to domesticate parrots. The trait humans were and are most interested in selecting for is how good a parrot is at talking to humans. In this counterfactual world, would parrots be capable of holding a conversation on or above the level of an average five-year-old human child?

22. Would an alien species of intelligent tortoises be able to create a technological civilization if the least intelligent healthy members of their species were as smart as John von Neumann?

23. Who would you rather have in control of America's nuclear weapons arsenal? A cunning criminal or a narcissistic imbecile?

24. Suppose that you had an IQ of 120 and an unconditional annual income of $100,000. If you could sell your IQ points for $100,000 per point (single payment), how many IQ points would you be willing to give up?

25. What scenario would be the scariest? A. ʻOumuamua returns; heading towards earth. B. Google shuts down its servers after an AI takeover. C. A breakthrough enables cheap and ubiquitous human memory extraction. D. A breakthrough gives China overwhelming military dominance. E. Profitable mass production of antimatter becomes feasible. F. Pakistan distributes its nuclear weapons among global Islamist groups.

26. Suppose we found definitive evidence for an alien megastructure 100 light-years from Earth. How would this influence you? For example, would this affect any projects or long-term plans you have? Would you become less worried about existential risks?

27. Which risks from artificial intelligence are likely to come first (assuming both risks are real)? A: Narrow AI, similar to GPT-3, which can output a blueprint for a superweapon (e.g. an extremely lethal bioweapon in the form of an airborne virus) or B: Strong AI that destroys us as a side-effect of an innocuous goal (paperclips maximizer).

28. Under the assumption that transformative AI (artificial general intelligence) will be created, what's the most likely route by which it will happen? A: Incremental improvements: It will happen in an evolutionary fashion. The improvement and combination of many different machine learning techniques, narrow AI approaches, and insights from neuroscience will at some point cross the human-level threshold. B: Paradigm shift: One or a very few conceptual breakthroughs will suddenly result in human or superhuman-level intelligence that is fundamentally distinct from narrow AI systems.

29. Suppose you could time travel to the year 2121 to a vacant place above the ground in your city. After 30 minutes, you, or any configuration of atoms your body has been converted to, will return to your previous position in the current year. You won't be able to take any technological artifacts with you that are not made of the atoms of your body. Would you do it?

30. You can choose the year and the amount of access to sex. Which option do you choose? A. 2021: same access as now. B. 2121: access to super-advanced biological sex robots you can freely design. C. 1921: access equivalent to a typical billionaire. D. 10,000 BC: Genghis Khan-level.

31. You are placed under solitary confinement for one year. But you're allowed to take 3 books with you. What books do you pick?

32. When you suddenly wake up from noise, does it feel as if you have consciously perceived the sound?

33. If you think about video games you played more than 20 years ago, do the graphics look much better in your mind than they actually did in reality?

34. How do insects that possess a wide field of view avoid damage to their photoreceptors from prolonged exposure to direct sunlight?

35. How come software engineering is almost completely unregulated despite being much more impactful than e.g. bridge engineering? Many states have food inspectors but not algorithm inspectors even though the latter influence billions of people.

36. You must choose one of the following two options, which one do you choose? Option1: Everyone's IQ except yours is set to 180 while yours is set to 80. Option 2: Everyone's IQ except yours is set to 80 while yours is set to 180.

37. Given widespread societal and political support, would Nigeria be able to build a modern aircraft carrier within less than 40 years and without significant technical support from other countries?

38. Would Hong Kong have been better off without British colonization?

39. Are Tibetans today better off than they would have been in the counterfactual scenario in which Tibet would have maintained its independence from China?

40. What would be the leading cause of death in a perfect world?

41. What percentage of your moral principles do you expect to have in common with an intergalactic empire of superintelligent alien slime mold?

42. Suppose an uncensorable anonymous communication network was created to which everyone would be granted irrevocable access at birth via brain implants that cannot be disabled. Everything would be allowed, from bioweapon blueprints to child porn. How would you feel about this?

43. What would the feasibility of warp drives mean for the Fermi paradox?

44. Suppose there was a free AI service that claims to be able to automate your social media presence. You test it for a few weeks and it turns out that you not only agree with everything it posts but it's actually a smarter, wittier version of yourself. Would you use the service?

45. If people truly believed that Jews outcompeted everyone in various intellectual tasks because of their cultural milieu, shouldn't we expect to see a trillion-dollar industry trying to figure out how to teach gentile parents to raise their kids in a similar way and see billionaires converting to Judaism?

46. What if the great filter is tail risk aversion preventing civilizations from reaching technological escape velocity?

47. Could offensive bioweapons research be disguised as pest control research? In other words, could people who legally design "bioweapons" for pests such as rats, ants, or mosquitos do the same for humans? Do we know who these people are and is their research subject to controls?

48. Do mental disorders like schizophrenia affect one's ability to think rationality in general or would it theoretically be possible to conclude that the hallucinations and delusions one experiences are more parsimoniously explained by mental illness? And what about hallucinogenic drugs. Are people actually capable of inferring that what they experience are just malfunctions of their brain?

49. There are hundreds of proofs of the Pythagorean theorem, some of which are much easier to understand than others. Does every theorem necessarily have more than one proof? How many? And could there be some way to formalize and sort proofs by their understandability?

50. Assume that there was a cheap and reliable way to reverse and prevent aging. It could make and keep you youthful and fit, both physically and mentally. How much would this change your current plans and behavior?

51. It's widely believed that regular exposure to consistently elevated sound levels (50 dB at night) has many negative health effects. Nocturnal rainforests have an ambient noise of 60-65 dB. How does this impact people living there? Do they sleep worse than people in quieter regions?

52. Suppose that western far-right parties started to ally with Muslims and campaign for their mass immigration in order to defeat secular liberalism. As a result, a majority of Muslims would start to vote for those parties. How would the political left react to this development?

53. Suppose that, for the next million years, the whole world became a caliphate in which women were forced to wear a burqa and all marriages would be arranged. Wouldn't this effectively replace sexual selection with selective breeding? And how would this affect human evolution?

54. What decrease in your income and the income of people like you would you be willing to accept in order to prevent or induce fully open borders in your country?

55. Thought experiment: Brazil, Peru, and Colombia decide to turn their share of the Amazon rainforest (83%) into a giant parking lot. Would you support outside military intervention if this was the only way to prevent this?

56. People from which of the following fields do on average have the best grasp of statistics? A. Bioinformatics / Genetics B. Actuarial science C. Hedge fund management D. Machine learning

57. What do people living in East Asia or places like Singapore think when they watch Western movies and see walls defaced with graffiti, parks full of dealers, and playgrounds soiled with drug needles?

58. Is it in theory technically feasible to somehow secure whole-brain emulations against abuse once it becomes easy and cheap to copy and run them? Could one e.g. give an emulated person a killswitch that allows it to terminate its own execution at any time? And could techniques such as homomorphic encryption and formal verification be used to make this killswitch unhackable?

Other questions I have asked but which are not my own:

• How many holes does a drinking straw have?

• You’re given a box with a set of dice in it. If you roll an even number, a person’s life is saved. If you roll an odd number, someone else will die. Each time you shake the box you get $10. Should you do it? (Source)

• “If you consider personally refusing to eat meat to be an effective use of your time and effort (to reduce animal suffering) then why don't you consider personally refusing to buy or use new computer parts an effective use of your time to reduce AI risk?” ― Roko Mijic

• Can we invent new states of matter as different from what came before as something like consciousness is from other states of matter? (Source)

• Do the ultimate fundamental laws of physics support structures capable of understanding those laws? (Source)

• You’ve made a pact with a friend that if they commit suicide, you will kill their cat (to disincentivize their suicide). imagine the friend has committed suicide, would you follow through and kill the cat? (Source)