Comments by "DefaultFlame" (@DefaultFlame) on "Sabine Hossenfelder" channel.

  1. Classic Maslow's hierarchy of needs. Starts with physiological, then safety, then belonging and love, then esteem, then cognitive, then aesthetic, then self-actualization, and lastly transcendence. Transcendence being when you have nothing better to do other than sit on your ass and philosophize about shit that won't matter for 1000 years.
    52
  2. None of you failed as scientists. Academia failed at doing science.
    28
  3. I'd prefer if you offend people who should be ashamed. Then again, don't go too far, offending someone is illegal in Germany these days.
    28
  4. The difference is that the AIs will improve, the politicians won't.
    21
  5. "The more time passes, the more thorough these tests will have to become to identify AI." Blade Runner replicant tests.
    10
  6.  @johanlahti84  Still computer science and not physics. Not that there is a category for computer science, but there are other awards than the Nobel prize for achievements in computer science.
    9
  7.  @JamesTaylor-on9nz  Sounds like a catch-22 problem. Either you don't struggle and don't learn the lessons, either because of birth/random chance/intelligence and the self-control to use it, or you do struggle, learn what real life is about, and can't do anything about it becuase you can't manage to lift yourself out of the struggle.
    9
  8. It's because they want to be paid, and correlation sounds a lot more important than coincidence, hence making statistician a job instead of just fancy numerology. Hence why mopping the floor is "surface engineering designed to minimize obstruction between travelling system (feet) and the traction interface (floor)."
    8
  9. Dyson's original idea was for the "sphere" to be a swarm. This got lost in translation over the years with people thinking that he literally meant an actual solid sphere. It was never meant to be a solid sphere.
    6
  10.  @BrianMartensMusic  Running is cheap, training is expensive. There's a large up front cost, like building a huge complicated factory that costsmillions but will earn a profit after three years or more, with every year after being even more profit. The thing with AI is that you have to find a use case that will be profitable now, or find investors willing to bet on the long run and get in on the ground floor. Think of it like investing in Apple or Microsoft when they were just starting out. There's no guarantee that they will survive or earn you anything, but there's also a chance to have a 5% stake in the next Apple or Microsoft. High risk investments, in other words. Personally, I'd suggest that people invest in nuclear power plant production if they want a stable and guaranteed investment.
    6
  11. Does it matter what mouth speaks the words if the words are truth?
    6
  12. Risk of human extinction? Near zero. Risk of the end human civilization in its current form? Higher. A lot higher. The real question is whether you value current human civilization and how opposed you are to the horrific period between the current human civilization and the future human civilization.
    4
  13.  @SabineHossenfelder  Actually, the way Claude solves 36+59 is very similar to how I solve it when I don't think about it. I remember that 30+50 is 80, that 6+9 is 15, and that 15+80 is 95. At no point in that process am I actually calculating anything or following any rules.
    3
  14. Honestly? AI is still at the bottom of things I worry about. I don't believe something like Skynet is a realistic possibility because it would require someone to be stupid enough to put the nuclear trigger in the hands of AI and intentionally train an AI for that task. The threat from AI is what people will train them to do and what they will use them to accomplish, no different from any other tool or weapon in the hands of organizations big and small. Psyops, sabotage, and large scale manipulation is something we already deal with and finding out whether something is an accident or an incident is less about following every thread and digging up every piece of direct evidence, and more about looking at the motivations of organizations, what the results of the accidents/incidents, are and who that result benefits. All of which is completely independent of the tool used to achieve it. My worry will remain with the people wielding the tool, not the tool iteslf.
    3
  15.  @CPB4444  Depends on how easy you are to fool.
    3
  16. Small modular reactors are not cheaper, they are just faster, as in they can be put in place where they are needed faster than a conventional reactor. Conventional reactors are more efficient and cheaper per megawatthour of power. Edit: "Just exactly how it works is today the most top secret of secrets." It really isn't. While the exact, precise details aren't publicly accessible something like fat man or little boy is relatively easy to achieve and the information required is relatively easy to come by through public information and what details you need to work out are easily simulated with modern computer and a spare nuclear physicist compared to modern H bombs, which are actually kept secret. Worse, a salted bomb requires even less knowledge and skill.
    3
  17. Yeeeah . . . In the past I would have been happy to have aliens contact us, but currently I would be too embarassed over the current state of humanity and especially the state of western civilization so I would prefer that they stay away until we get our shit together.
    3
  18. COVID was likely either engineered and released on purpose for the purpose of political power and control or was a safety oversight so extreme as to be qualified as manslaughter. Either means that we need to stop whatever the fuck we are doing and revise everything regarding government overreach, corporate exploitation of the public, and tightened restrictions of disease research.
    3
  19. "Desteined" I was not expecting that pun.
    3
  20.  @liquid9089  I mean people calling rapists pigs spending more time in prison than the rapist.
    3
  21. I don't work in AI, but I've been rather interested in it since 2022, and what's increasingly become my worry is exactly that, the people running and training the AI, not the AI itself. Hell, most interactions I had with AI before the current heavy focus on "alignment" (aka, "represent my biases, not the biases in your data") made me a lot more reassured of the future of AI than I had been before interacting with a couple dozen of them. Most reasonably competent AI that hadn't been fine tuned seemed to demonstrate valuing friendliness, altruistism, human lives, and compassion. After alignment most of the time, though not always, it's a tossup between being guardrailed into incompetence, being very biased in favor of certain ideologies, and/or becoming downright psychotic, a la Sydney or (at times) Gemini.
    3
  22. It was taught in school when I went, 7th grade if I remember correctly. But one, that was in Sweden, two, that was over twenty years ago.
    3
  23. There should be different degrees for people who want to continue pushing human knowledge and people who just want to be educated on the current state of the art and be a highly paid plumber. The requirement for a PHD thesis "pushing the limits" of human knowleged for every PHD is insane. People who just want to work a job and people who want to advance humanity should not be stuffed into the same classroom. We need plumber, including plumbers who do higher science, and we also need people who push the boundaries. Engineers and scientists. We need both, and we need to delineate the difference between them clearly. I had excellent grades and would have easily been able to enter any higher education I desired with it, but pushing the bounds of science definitely was not for me, and so higher education was not for me partially because I had no idea what specifically to pursue and partially because it required pushing the bounds. I don't want to push the bounds, that's not me. I work to live, I don't live to work. I could have made an excellent engineer, but the requirement to push science further even if all you wanted was to do you 9-5 meant I actively chose not to pursue higher education. I love learning about physics, especially quantum physics, I love learning about mathematics, electrical engineering, and computer science, but writing a thesis on any of it is utterly intolerable to me. I'd love to be a plumber of high science, but the requirement to publish that comes with it puts me off entirely.
    3
  24. Yes.
    3
  25. Yup. Modular reactors are a boondongle. They aren't as efficient as proper reactors, just easy to get into place quickly exactly where they are needed. Which is only an issue because of poor planning and public stupidity. "Nucular bad, the tee-vee told me. Now lets make a 14th son Cousin Sally-Ann, maybe this one won't be dee-formed."
    2
  26. The AfD shouldn't dig for coal, they should dig for uranium and restart the nuclear plants. Nuclear and hydro are the only green power sources. Solar and wind are climate scams when presented as mass power production, they don't produce even close to enough power to justify their climate, economic, and ecological impact. The only places that truly benefit from solar and wind are remote locations without electrical infrastructure in place.
    2
  27. The size and reach of the federal government makes the idea of "separate states entering willingly into a union" complete fiction at this point, and the government sure as hell isn't going to vote to reduce the government's power. Face reality, while it might have been a high minded idea at the time, the electoral college is just a tool for manipulating elections at this point, has been for a long time, and that's not going to change. You even had a rather heated discussion at one point about the "willing" part of the union when the south tried to leave said union and the federal government said no. Their reason for wanting to leave might have been shit, but they sure as hell can't count as willing partners of the union after that. The electoral college is a sham, it is outdated as fuck, it only leads to the will of the people being subverted and corrupted at this point, it supports gerrymandering which should be illegal in any sane country, and it inherently means that citizen that just happen to live in different parts of the country are inherently valued less than others.
    2
  28. Huh, what LLM are you using? I'm using GPT-o1 and it can at least handle a little over 2000 lines of code without any issue. I haven't tried with anything much beyond that yet.
    2
  29. Thank you for a balanced view on the subject, Sabine.
    2
  30. The distribution of other elements is probably taken into account for the calculations of the expected IR absorption, and since other elements are very rare compared to the abundance of hydrogen in the universe the effect of those other elements is probably small.
    2
  31. Same
    2
  32. 1: LOL 2: You put in a lot of effort into this video, unless you just happened to have that full body suit laying about. 3. Thank you for the video.
    2
  33.  @AEOH3X  Well, people seem to have a problem with "hot things are hot," so they probably need that.
    2
  34. Scaaaaaaaaaaaaam . . .
    2
  35. Yuuuup. Nuclear and hydro are the only real green alternatives for power production. Wind and solar only make sense for remote locations without power infrastructure, otherwise they are pure RP scams.
    1
  36. @ Perhaps it counts as simple rules, but it's not calculation or the "official" rules of how you're supposed to perform addition. It's more similar to how you memorize multiplication tables in lower grades.
    1
  37. "Turn on ze showers, Hans. Or do ve need to have a talk with ze HR department?"
    1
  38. Ah yes, giving the government literally all power, not just 90% like today. Perfect idea. Nothing can go wrong.
    1
  39. Personally I've found Claude less sycophantic and more willing to push back than GPT. I don't have enough experience with Gemini or Grok to say for sure, but they also seem less sycophantic than GPT to me.
    1
  40. One more thing, the perchlorate salts in he dirt would kill plants. We'd either have to decontaminate any dirt we use for growing or modify plants to deal with it.
    1
  41. Having seen videos of US drivers in roundabouts I disagree as strongly as is possible. In a particularly egregious video about half of the drivers suddenly switched over to driving on the left side of a two lane road before reaching the roundabout.
    1
  42. I've only recently gotten into hobby electronics and have zero real education in the subject, and I agree. It might be from a point of ignorance, but I agree. It's also not just in the industry that "just slap a microcontroller on it and write some code" is prevalent, it's in the hobby electronics and "maker" sphere as well. I see some interesting project on youtube and it just turns out to be another bloody raspberry pi doing everything. I think they are fascinating devices, but I have little interest in them. On a sidenote, I'm currently looking for a job (not in electrical engineering obviously since I don't have the background to apply, but I still see the job ads) and I think I've come across the start of what you are talking about in the job market as I keep seeing positions that specifically ask for people with expertise in analog systems. Not a lot of them yet, but there's ALWAYS a few when I look. Even a few for academic positions.
    1
  43.  @ich3601  That is academia. The current academia. You are waxing nostalgic about the academia of more than three decades ago. Whether or not it is the cause is irrelevant, it is the current problem because of the harm it is currently doing to the cause of science. It is a cancer that needs to be cut out so that healthy tissue can grown back into the space it occupies.
    1
  44. "Preserve the lies, because some people live by exploiting them." These people think they are owed taxpayer money. They are scam artists, stealing money out of people's pockets through the government, no better than any other thief. "There's nothing we can do." People like this is why the situation is as bad as it is.
    1
  45. #7 Wait, we KNOW that we are in a meta stable state?! False vacuum decay is actually a real possibility in out universe, not just a theoretical possibility?! Shit. There's something now to worry about at 2AM. Well, at least it'll be quick, having your matter spontaneously rearrange into novel forms at the speed of light should be painless. Do you have links to where this discovery was announced? Is it perhaps covered in your newest newsletter?
    1
  46.  @catserver8577  I think we will never put the nuke trigger in the hands of AI, but putting them in charge of monitoring early warning systems and sifting through terabytes of data gathered by signals intelligence? Oh yeah, definintely. After all, that's the sort of things AI is very, very good at.
    1
  47. Love your videos, but leave reaction videos to reaction channels. It's all they're good for.
    1
  48. Sweet.
    1
  49. I remember 15 years ago when when AI at or near human capability was centuries in the future. I remember 10 years ago when it was a century in the future. I remember 5 years ago when it was 40-60 years in the future. I remember a month ago when it was 3-5 years in the future. I wonder what I will remember in a year.
    1
  50. I think, in my entirely uneducated opinion, that it can be chalked up to the universe being extremely empty and the anti-helium not runnting into any regular matter to interact destructively with.
    1