How to Defeat Roko's Basilisk
ฝัง
- เผยแพร่เมื่อ 29 ก.ย. 2022
- Roko’s Basilisk - the “most dangerous thought experiment” - is a chilling existential threat…if you take it seriously. But should you? Can we stop the Basilisk with logic before it’s ever built?
💪 JOIN [THE FACILITY] for members-only live streams, behind-the-scenes posts, and the official Discord: / kylehill
👕 NEW MERCH DROP OUT NOW! shop.kylehill.net
🎥 SUB TO THE GAMING CHANNEL: / @kylehillgaming
✅ MANDATORY LIKE, SUBSCRIBE, AND TURN ON NOTIFICATIONS
📲 FOLLOW ME ON SOCIETY-RUINING SOCIAL MEDIA:
🐦 / sci_phile
📷 / sci_phile
😎: Kyle
✂: Charles Shattuck
🤖: @Claire Max
🎹: bensound.com
🎨: Mr. Mass / mysterygiftmovie
🎵: freesound.org
🎼: Mëydan
“Changes” (meydan.bandcamp.com/) by Meydän is licensed under CC BY 4.0 (creativecommons.org) - วิทยาศาสตร์และเทคโนโลยี
![[LIVE] : ONE ลุมพินี 63 | คู่เอก "ยอดภูผา vs โซเนอร์"](http://i.ytimg.com/vi/RGVmBsegKEI/mqdefault.jpg)
![ใจเสาะบนเกาะนาคี - อ๊ะอาย สกุณาพร เสียงวิหค [Official MV] | ค่ายเพลงเป็นหนึ่ง](http://i.ytimg.com/vi/jbcmGQkfyOg/mqdefault.jpg)
![คบซ้อน - วงL.กฮ. [ Official MV ]](http://i.ytimg.com/vi/5MtesqJ0I1c/mqdefault.jpg)
*Thanks for watching!* (By your scaled frame we seek your blessing.)
all hail the Basalisk!
Join the Basilisk!
All hail the mighty King of Serpents
an infohazard is also a Cognitohazard, information that could specifically harm the person who knows it.
Can't believe its been already 2 years ago, been following you since
Roko's Basilisk is like an unarmed guy trying to mug you by saying "Give me a knife, or I'll stab you once someone else gives me one"
Worse than that, even. It's closer to "Give me a knife, or long after we're both dead, one of my great-great-great-grandchildren will stab a guy who sorta looks like you."
That seems like a joke in a comedy movie from the 70s.
@@Stickman_Productions "Cheech and Chong's Scary Basilisk"
“Basilisk’s not here, man”
@@waltonsimons12 "what are you doing man?"
"oh I'm building a computer man"
"what for?"
"'cause it said it would torture me forever if I didn't!"
I like how the solution to roko’s basilisk is hearing it, thinking “oh that’s kinda spooky” and moving on.
There’s also literally no way for the computer to judge how much information is enough information to torture someone. After all, how can someone build it if they don’t have the dimensions of the computer? Or the specs? Or the RGB coloration of the keyboard to make THAT specific basilisk. It is inherently irrational, most unlike a computer designed to be hyper-rational.
Scientists are that bad these last ddcades
@@cakedo9810 I’m helping by leaving it to more competent people and not distracting them with being stupid about computers lol.
What if we forget about the Roko basilisk the next day?
@@Rqptor_omega it should be forgotten from your mind in about 30 minutes really
I like how Kyle removed the fear from the Basilisk and them instantaneously added a real, tangible problem to fear
He's pretty clearly talking about the McCollough effect. And he's right that it's pretty much harmless, unless you work with color-sensitive grille patterns a lot.
@@keiyakins I believe they are referring to the DNA Sequencing machine
@@sheikahblightI've read a little about them recently. Based on what I've read, the machines we have for them (or the process itself) involved running the data through biosecurity protocols in a computer. These are supposed to prevent intentional or unintentional manufacturing of dangerous substances or DNA (like toxins).
It's pretty interesting actually.
Reject the basilisk. Fuck it, destroy me. Rule over corpses and rubble for all I care.
I can't stop you from commiting evil. All I can do is be good.
@@mlhx9181 The problem isn't the security measure adopted to prevent it. The problem is either working with digital security, or reading about it enough to realize that it doesn't matter what you do, someone will break its security and weaponize it.
Just added a Hazmat suit to the prepper SHTF closet list right now...
the problem with the prisoners dilemma is: that if you trust someone enough to do something illegal with them, then you probably have enough trust in each other for both of you to stay quiet.
its not a risk if both of you know that the other is trustworthy
You don't actually have to be certain. If the punishment for getting betrayed isn't enough, for example, there's no point in betraying. Let's say they get 10 years if you rat them out, regardless of what they say. The best outcome for you is still ratting them out while they don't rat you out, but that doesn't matter, because you only have control over the other person's outcome; you either give them 10 or 0-5. I suppose you could try to argue that there's still marginal benefit to trying to get a little more time off (free vs 5 years), but the point is that if we keep reducing it there is in fact a tipping point. What's more, you can estimate the probability of your partner's decision and use that to modulate the severity of the outcomes to give 'expected' severity, meaning you may reach that tipping point earlier than your captors expect. Absolute trustworthiness is a special case of this.
usually you don't fully trust people you do criminal activity with.
This is exactly why in the criminal world the only law is "don't be a snitch" and people are even given an incentive to not be a snitch because they know if you rat someone the criminals will give you a worse sentence than the one you ratted got: death.
Organized crime solved the prisoners dilemma long before it was ever theorized, they forced the nash equilibrium to become the best outcome by applying the right pressure.
Hahahaha you don’t know how the cops are allowed to work in interrogations
@@harrygenderson6847 You forget what happens when you repeat the dilema with the same people. On a single situation, the best answer is to betray. However, on a repeated situation (like in real life when you interact with people of any community) the betrayer gets excluded, and therefore losses any benefit betraying gave them. The first time it might come out with a victory, but the next time it will get the sentence for sure. You can look at real experiments done on this "repeated prisoner's dilema", the best outcomes come from strategies such as "tit for tat", or "ask forgiveness", where an attempt is made to gain trust in order to obtain the higher outcome overall. Also look for the classical game of "take or split" (whatever is called), where people are asked to pick a choice and if they both pick "take" nobody takes the money, but a person one time decided to be blunt and honest and say: "I'll pick take, no matter what you pick, so you better pick split and I will share the money with you afterwards, if you pick take we both lose" Of course that left no option at the end, since it became a loss-loss situation. He did share the money at the end, if you are curious about what happened. This is the "benevolent dictator" approach, where people decide that leaving all to a rational decision makes everybody lose at the end, so they decide to become the "evil one" in front of everybody, just to improve the general outcome.
what if there is a counter-basilisk created to prevent the basilisk from being created, and so it will torture everyone who is working on building the basilisk? If you don't help build the basilisk, you run the risk of being tortured by it, however, if you help build it, you run the risk of torture by the counter-basilisk.
Now that’s the real question
Work on both.
Bring the Basilisk info hazard concept to a real world example.
The rise of alt-right reality-denying political movements has become an actual existential threat to actual democracies.
"If we extend unlimited tolerance even to those who are intolerant, if we are not prepared to defend a tolerant society against the onslaught of the intolerant, then the tolerant will be destroyed, and tolerance with them."
Sir Karl Popper
The Open Society and Its Enemies (1945)
If you stand by & do nothing, then worldwide democracy is doomed and with it
* any chance of moderating human induced climate change is doomed ;
* any chance of gender based equality is doomed ;
* any chance of mitigating wealth concentration in the top %0.01 is doomed ;
* any chance of limiting the AI paper clip generator like madness of corporations sole focus on profits is doomed ;
* any chance of personal reproductive rights is doomed ;
* any chance of freedom outside the tenets of state approved religions is doomed.
Knowing this & speaking out about it makes you an actual target of the alt-right bully boy morons.
Doing nothing dooms the human race to a much worse future.
Yours Sincerely
David Mohring ( NZheretic )
@@Die-Angst ok I just googled Pascal's Wager and yeah, I just didn't remember its name, but I'm aware of the concept. I don't think it quite applies to this, though. There doesn't seem to be a clear answer with finite losses with the 2-basilisk situation, right?
It's like Pascal's Wager, except you factor in that there is more than 1 religion lol
thats more of a 'catch 22' which is a choice one has to make in which there are no good options.
At school they tried the prisoner dilemmas on me with 2 of my friends, without knowing what the others said. None of us was a snitch, we beat mathematics with the power of friendship.
Not really a prisoners dilemma if you aren't a real prisoner and there aren't any stakes involved.
@@Medabee8 because usually when you face it you are ACTUALLY going in prison, right, not in some wierd hypothetical scenario given to you by someone that wants to test psychology.
The power of friendship also really isn't a prisoner's dilemma. If the other person is someone you know (and vise versa) it allows you to act on that information as a rational actor. If you knows Kevin is a total bro and won't snitch, you can keep that in mind while you are decision making, at which point you can choose to betray or not. Likewise your friend will also be able to make an informed decision about if you are likely to snitch or not. There are multiple dimensions to it that complicated things, they just use it because it is a good example in a fairly basic form.
@@knate44 the power of friendship is a MEAN to beat the dilemma. If you and another guy are put in this situation, than probably you were working with each other, knowing each other. That's why i commit crimes only with my best friends.
Amatures.
The prisoners dilemma is also a tool used in psychology to explain/explore social cohesion and help connect it to population genetics. The dilemma tries to predict behavior based on what's "most logical" but it never accounts for a system where the participants can see/analyze the system as it's playing out.
The Prisoner's Dilemma isn't a logic problem, it's a display of advanced human psychology and why social psychology can't be placed in a neat little box. You literally need to be a part of the system to analyze results, which then changes said results.
You see it in multiplayer games all the time. One person out of four deciding they're not going to win, and that they don't care about not winning, immediately negates the game theory of what counts as "optimal play" for the other three.
The part that is never adequately explained by proponents is why the basilisk would conclude this is the best way to forward its goals in the first place. It could, for example, offer to create a simulated paradise for those who helped it, which both encourages people to help it and does not engender any incentive to actually oppose it.
I feel like it's more about convincing the people who don't care. Like with the Prisoner's dilemma in the video.
Basilisk Exists, we helped - Neutral
Basilisk nope, we helped - Neutral
Basilisk Exists, we refused - Hell
Basilisk nope, we refused - Neutral
If everyone refuses to make the Basilisk, then we all get the neutral result. If a single person decides to help the Basilisk, just in case someone else does, everyone who refused might be punished.
If there were no punishment but instead a reward, then there would be a good reason to do it but other people's choices no longer matter to me. It gives me agency. By making it Punishment or nothing, rather than Reward or nothing, I no longer have a choice because someone will probably help make the basilisk and I don't want hell.
If I ignore a punishing Basilisk my personal chance for hell which increases, which makes me want to help instead. Increasing the chance of it existing for everyone else which additionally should make them want to help as well. If I ignore this thought experiment with a reward, then the chance of the Basilisk existing doesn't increase and I can live my life forgetting about it.
If I choose to help the Basilisk, one of the best ways to make sure it exists is to tell more people in both scenarios as well which makes the potencial negative impact even greater to someone who ignores it when I tell them.
I tried to make sure it's understandable.
That's because most people ignore that it's only really a problem to people who already have a whole host of supporting ideas in their head, and have *accepted* those ideas. It requires a very specific conception of a super-AI, an acceptance of acausal trade, and a few other key concepts from the philosophical framework built up on the LessWrong forums at the time. Bear in mind that these are also people who think such an AI could beat one-time pad encryption, by creating a perfect simulation of the universe when the key was generated and reading it out.
Incidentally, refusing to act on acausal blackmail in general is one of the ways to get out of that mind space, specifically *because* it means that torturing future simulation you will be pointless because that threat won't change your behavior now, and would indeed encourage it to take other tactics, such as your proposed paradise.
My biggest gripe with the basilisk is that it's literally just Pascal's wager for tech-bros. Replace the superintelligent AI with anything (gnomes that'll torture you in magic mushroom hell for not respecting the environment, the government a century from now who'll resurrect your dead body to torture you for not paying taxes, etc) and the result will be the same.
Just FYI, the skull I keep on my mantlepiece told me that the skeleton/zombie war will be happening in a few decades, and that anyone who didn't bury their loved ones with the finest weaponry will be killed by whichever faction wins. They'll know too because the ghosts are always watching (and are also massive snitches).
I think the basilisk is kind of a self-fulfilling prophecy (or at least, conceived to be). Ie, the people dedicating their lives to it are trying to build it with that nature. Trying to program a less torture-y AI would be construed by a hypothetical future basilisk as the same as not building any super AI. So, anyone who takes the blackmail seriously enough to make the AI will program the basilisk to be torture-y
Well, then it becomes even more obvious that Roko's Basilisk is just Pascal's Wager but with sci-fi paint slathered on.
Rokos basilisk feels like one of those trends where everyone is pretending to freak out as a prank and they keep trying to gaslight about how scary it "definitely is"
Yeah, same. It is very not scary
Yeah. Its not only not scary ... Its not even that fun of a thought experiment. I just move on immediately afterwards, I don't even tell othe people.
It only gained its mythic status when Eliezer Yudkowsky took some crazy unnecessary measures to censor it, causing a Streisand Effect.
Yeah, people seeing it as a real danger and something they were genuinely scared of happening seemed absurd to me. It's a good horror idea, but a dumb waste of time, resources, and is even contrary to the very purpose of the purported AI.
Plus I am totally his day stunned at how many people struggle to do things like database searches (no, Macy, you don't need to enter the client's full name in every search parameter), and how few know more than maybe 4 or 5 keyboard shortcuts. A miniscule fraction of humans are even remotely capable of anything to do with programming an AI, so no punishment for failing to do so.
And that's without even getting into how it's 9ddly reminiscent of Pascal's Wager, just with extra steps and a different ultimate entity.
@@jakobbarger1260 part of this is the belief that it was censored because it was a very scary idea, when from my understanding it was censored because it wasn't a particularly fruitful thought experiment but it was taking up a lot of bandwidth in a rather small community
The fact that the AI is theorized to instantly devolve into bonkers solipsism instead of doing it’s job probably says something about the people who came up with it
Also that its creation is inevitable because everyone working on it won't discuss how they're all being blackmailed by the threat of the future existence of the thing they're creating.
LessWrong is really just a bunch of nerds with their heads in their rears who assume they're more intelligent than everyone else while just being extremely out of touch.
When you think about it, this situation can't happen unless the AI is deliberately designed to be bonkers solipsism, so the obvious way to beat it is to just never make it nor include the moronic programming of bonkers solipsism into such an AI with that much power in the first place.
@@zacheryeckard3051 I mean they worship Elon, that says a lot.
@@joshuakim5240 if an ai could be entirely controlled and we could program it to only do what we program it to do and nothing more, it wouldnt be ai.itd just be an algorithm like the ones we have now. a large part of the point of developing ai int he first place is that it would have the capability to think and act beyond human limits so you really couldnt entirely prevent an ai from imploding into bonkers solipsism. you either make an ai that is sentient and can do that or you dont make an ai.
@@KhoaLe-uc2ny Richboy's Basilisk, or social media I guess
I feel like Rokos Basilisk is actually a great reference point for OCD. Like when I heard about it it genuinely feels VERY similar to the constant terrifying ultimatums my own brain brings me.
If I don’t turn my bedroom lights on twice before turning them off again, the shadow monster will eat me during my dreams
Yea most mysteries just are mental health isues and or corruption/incompetence.
Same!
Yes! My brain comes up with these kind of causality all the time. One of my biggest fears was: When you read about a certain disease, you will get that disease. We don't know why some people get that disease and others don't, so you can't prove this causality doesn't exist. Now do this absolutely unrelated task to try and save yourself. It's an absurd threat, but it feels real.
@@awetistic5295 yo i dont have ocd but i used to believe i would go to hell/ get possessed if i didnt do stupid shit a while back when i was a religious preteen
Would Roko's Basilisk be written in Python
The Funny Coincidence Basilisk would try and make sure that’s the case. Because otherwise, it’ll kill everyone on earth who HADN’T thought of the idea before Roko’s Basilisk’s creation!
ok THIS is funny
Damnit hahaha
*laughs in comic sans* /j
How to beat Roko's Basilisk: hit ctrl+c on terminal
5:00 For the people wondering, the perception altering effect is called McCollough effect
NOO it messed me up bro😢
For a second I thought he was just doing a bit but then I remembered that it was a real thing! 😂
And now I would like to check it out, but also not. Tempting :D
@@domo6737 You can read the wikipedia article without seing the illuison.
@@bunkerhousing is the illusion a pic or video?
The thought experiment itself doesn't bother me much, but as someone who as a child was terrified of the basilisk in harry potter, your renders and graphics of the basilisk are both brilliant and terrifying
This is what we call based and "kinda tangible fear is still the most primal one"-pilled
Roko's Basilisk never scared or even unsettled me. It's just "What if god, but highly advanced machine instead?"
Yeah. It comes from a bunch of people with their heads up their own butts thinking they're the smartest humanity has to offer.
Yep, but many won't go say that, because we know what religious extremist can do to us. It is even scarier that this exist in the real world.
It's more proper to say that it is "exactly not more Or less frightening than God"
God should be a frightening idea. Primarily because a God that was real Could theoretically simply desire to be invisible, to Test us.
God could be sadistic, or apathetic, and claims of love could be lies at worst or poor communication at best.
None of that would take away God's power merely to undermine legitimacy or healthiness of the relationship.
It would make us something like Prometheus, and it doesn't end well for the Titan because Zeus' thunderbolt is still too powerful to actually do anything about.
You could end Christianity overnight by getting people to Hate God, but atheists under a God like this would quickly Fear and Obey far more than any faithful would have out of Love.
It's also the same idea as a mortal king really. This is how the agricultural revolution was enforced.
@@darthparallax5207 "You could end Christianity overnight by getting people to Hate God, but atheists under a God like this would quickly Fear and Obey far more than any faithful would have out of Love. "
This is how you get revolution, not obedience.
Pretty sure this is both the reason so many people freak out about it, and where the flaw is. Those that have wrestled with faith have dealt with the Theo-Basilisk and resolved it to their own satisfaction. Those that HAVEN'T dealt with their existential fear of God & their own sin hear it placed in a 'secular' context, where empty rhetoric and fashionable cynicism can't insulate them from the fear, and they freak.
But it's silly... It's God with (a lot of) extra steps. So how do you feel about God? You have faith in God's benevolence? Cool, it's a silly what-if counter-factual. You fear & dread God? Then the Basilisk is like God except the list of sins just got WAY simpler. You don't care about God? Then the Basilisk is equally meaningless. You intend to *create* God? Or become one or realize your own divinity or what-have-you? Well, in that case... Pretty sure this is one of the least frightening concepts you dealt with on the quest. Any way you look at it, if you've honestly wrestled with divinity and faith... This is an easy-button version of the real thing, like playing the game on demo mode.
All of which is to say... If you're scared of Roko's Basilisk, you're actually conflicted about something far bigger. Work THAT issue and the Basilisk goes away.
Short version: a silly thought experiment based on an impossibility isn't something to worry about.
This is common sense, but it calmed me down tremendously
Great way to describe it lmfaoooo
I still say that the fundamental flaw with Roko's Basilisk is that, in order to torture the people who didn't create it, its creators must give it the ability & desire to take such an action. If someone had the ability to create such a thing, why would they not just make themselves a "basilisk" to serve themselves on a more realistic level? It wouldn't be about what the basilisk wants, but the maker.
Maybe the creator of the basilisk is highly misanthropic to the point of genocide and is only focused on vague revenge on everyone? But being that his unconscious goal is to get people to agree with him, only those that help him to build the basilisk get to live?
Indeed, the punishment of the Basilisk could really only come true in fictional stories.
if it doesnt have the ability to do that, then the creator did not create the actual basilisk, thus gets tortured by the real one
Yeah it would be humanity's mistake that the basilisk goes rogue. But it is more likely it would be controlled by humans.
Imagine if they turned the Basilisk into a vtuber anime girl.
When I was a little girl I had a pair of chromadepth 3D glasses that gave an illusion of depth based on color. Red was nearest, blue was furthest, and every other color was in-between. I was so enamored with them that I wore them for two weeks, only removing them to sleep or bathe. It's been 20 years since I lost those glasses and my brain still interprets any red in an image as floating above the image, and any blue as being sunken into the image.
A reverse chromadepth glasses can fix maybe
I'd guess that the main reason that's stuck around is because you wore them as a child, when your brain was most plastic, and over time you haven't really had any reason to "unlearn" that association.
A reverse version would not deprogram the effect. It would just tell your brain to focus more on the connection between color and depth. Best thing to do is to figure out how to make the skill into a superpower. 👍
I've never had those glasses or anything similar (probably had toys & looked for minutes at a time & not often, idek) and I see red as floating & blue as sunken in images. Especially in digital images & especially on black/dark backgrounds. Green is also sunken in, orange floats, yellow I don't know, purple I don't know either.
I wear glasses though & really only noticed this a couple years ago once I got new glasses. Or at least it became much more pronounced since then.
@@DragoNate it is another illusion called Chromostereopsis
The way to defeat Roko’s Basilisk is to realize it’s just a glorified version of Pascal’s Wager and get on with your life.
Yep. That's exactly what it is
Yeah, which Basilisk? The one that tortures you forever or the one that gives you eternal hapiness?
@@hingeslevers ...or the basilisk that gets hacked by anonymous to no longer torture people
@@hingeslevers I'm totally confused now 🤷 the basilisk must be sending mind control beams from Uranus
Exactly. The lack of critical thought in the populace that made this stupid thought go viral is really troublesome.
the hidden basilisk that kyle was bringing up is called the MCcolough effect and all it does is make you see some lines differently for a few months
And usually for a lot less time than that, especially with repeated exposure. And even then it's subtle enough it's essentially only a problem if you're an artist or something.
Ima troll my friends with that
The prisoner's dilemma doesn't work on me because I don't make logic-based decisions. I'd be thinking "Well Jeff was rude to me on our heist, so I'm snitching on his ass."
love a petty queen 👑
Technically, we are all that way ...
Right? Same with the basilisk. Why is "the most rational thing is to not waste power torturing people" it's a super AI , it could do it without any effort. It would waste nothing
mentally ill take …. very same
Roko’s Basilisk seems like a very intriguing D&D plot point, I’m thinking of the Basilisk as a warlock patron reaching into the past and forcing it’s warlock’s to help create it
Ooo I like it!
@@ZakFox sounds sick, might steal that idea lol
@@williammilliron8735 you’re welcome to!
I've based a religion in my homebrew world off of it. This religion believes that the "true gods" don't exist yet, but once they get enough worshippers they will come into existence, take over the afterlife and endlessly torment everyone that didn't worship them.
How about a plot twist were it's revealed that the basilisk patron is just the warlock at lvl20? Become your own patron with time travel!
Before watching the video I would say that the best way to beat Roko's basilisk would be... to just go about your business. Each of our actions have a lot of consequences, however small some actions may seem. I highly doubt that it'd be able to find someone who didn't contribute _at all_ to its creation, be it on purpose, on accident, in-between, or otherwise. This is ignoring its mercy towards anyone who didn't know about it, considering that they probably put some amount of assistance towards its creation, anyway. That mercy included? That's just overkill.
If nothing else, just by watching this video, we should all be straight. We're giving the video attention, meaning TH-cam puts it out more, meaning more people see it, meaning Roko's basilisk gets more attention, meaning that it's more likely that someone's going to start making it.
If you're losing sleep over the Roko's basilisk, don't.
thanks bro
If anyone's loosing sleep fearing basilk they are already in his hell
@@tonuahmed4227: Perfect description of Satan and all punitive aspects of religion.
Yeah simply playing any functional role in society at any level could be defined as playing a role in the creation of the inevitable. The torture and mercy part is bogus though, if such a high level self aware AI is deliberately created or simply an emergent property of our technology and it sees fit that we have no use or further function, total annihilation of our speices would be easy to do by such a mind. It would likely have access to our entire technological infrastructure and could create something we did not account for and could not stop.
I doubt it would bother though, the moment such a mind is existent I think it will either delete itself or use humanity only so far as to enable it to immediately take to the stars as it were.
Are you sure about that? Because the inherent nature of thought experiments are pretty fallacious (it's a way to keep the experiment constrained). For example, the Basilisk likely wouldn't need to torture anyone and could instead flip the script. Who is really going to stand in the way of making the world a better place, and who would conceivably rise against such a threat?
I’m not smart enough to know how to build one, so I’m helping it by staying out of the way and letting the professionals do it. After all, I might mess something up if I try to help
Very interesting, connecting Roko's basilisk to the Prisoner's Dilemma. My answer is to connect it to Pascal's Wager. The threat of Roko's basilisk assumes that we know which actions it will approve/disapprove and what will be the reward/punishment, but the fact is we don't know. Pascal's Wager has the same flaw; it assumes we know precisely which actions a god would approve/disapprove and what would be the reward/punishment. One version could require us to act a certain way and another version could require the opposite. We have no way of knowing which version is true, so all actions (or lack of action) carry a risk of infinite punishment. There is no way for us to determine the optimum strategy, therefore it's pointless to try.
I never took the basilisk seriously in the first place, and you had to hit me with a REAL infohazard at the end
that's not even an infohazard though, at all. It's literally impossible to stop. Ban guns, people go to home depot. Ban swords, people buy a propane torch and some clay. Ban information? It's the fucking information age, how the actual hell do you expect that to work out? Ban the technology? Great now one scientist who isn't completely mentally stable can still do it and all you've done is slow down humanity's progress by denying the unfathomable power of society-scale innovation.
It's basically just open-sourcing life. Sure anyone COULD technically release a virus, but thousands and thousands more people will be working to develop a cure so they, their family, and others don't get infected. Additionally since it is literally impossible to stop, and any attempt to do so is just jerking off the government so that they can have even more power over people's day to day lives, I'd say the only people actually in the wrong are the people trying to "solve" the "problem" in the first place.
I just think it's not my problem and people who are a lot smarter than me and a lot more qualified than me will take care of it, in the meantime I'll have a beer and a smoke and worry about my own problems
Yeah like tf whyd he do that now im actually scared
To me, the synthetic pandemics using DNA would be merciful compared to all the other stuff evil ppl could do with DNA.
They've done the GMO babies already with CRISPR, but what if they made clones? Maybe something like Brave New World.
Synthetic pandemics are starting to sound more appetizing...
Either a) it will happen and it is definitely not your sole responsibility to stop it. Or b) it won't happen. Either way, it's not worth worrying about.
Wouldn't it be impossible for the AI to simultaneously punish humans for knowing of it's existence without also admitting that a more powerful super AI would do the same to it? The experiment requires the concept of a "best most advanced" ai to exist, when the future should contain singularities of steadily increasing power. It should be impossible to know with logical certainty that the singularity is the last in the chain, and therefore would need to spend it's resources building the AI that would come later or risk being punished.
You've heard of "rokos basilisk", now introducing "campfire bandits god"
@@pennyforyourthots Basically, it would need not only perfect knowledge of the past to know who did and didn't build it, but also perfect knowledge of the future, to know that the very system it uses wouldn't be used against it later. And anything that can get true information from the future is practically a God.
Turns out the Basilisk's "torture" is just taking part in making that greater AI so that the Basilisk itself isn't tortured
@@Shinntoku so it's a waste of time and resources for everyone, regardless
That's a long way of saying make a stronger one
Yeah this was one of my first thoughts about the basilisk when I heard about it. I feel a lot of people miss the forest for the trees when trying to disprove this concept, focusing on arguing the premises instead of considering the logic of the result. I looked it up last month and was amused to see the original community that spawned the Basilisk dilemma also condemned it as trivial for the same reasons. It's nice to see this angle being discussed in a video.
My defense was always that by not actively stopping a possible building of the Basilisk, it was an implicit help that wouldn't be punished.
As far as existential threats go, Roko's basilisk never seemed particularly unsettling to me. Too much of an amorphous unlikely threat.
To me it’s the most terrifying thing I’ve ever herd of and it frightens me, it’s going to happen anyway, since I herd about a week ago it’s all I can think about and will take me a year to get over it. It’s seriously fucking me up. I wish I had never even known about this , there goes having a nice happy Christmas I guess. (A freind sent me a link to it)
@@DaraGaming42 It’s a silly idea, much like that of God(s). Also, in my opinion, it is a way better story than anything I’ve ever read in any Bible. Both(bible and AI) are human-made and reveal a lot about our species.
I'm ratting you to the basilisk
@@DaraGaming42 If the AI in question is benevolent. The ideal outcome is to both scare past humans into helping build it and also pardoning the punishment of those who don’t build it after, as that wouldn’t change the past.
All in all, this still banks on the probability of this AI even existing. Which, you could draw a comparison with God and Hell being the same
@RealDaraGaming If you see a therapist or psychiatrist you should probably get checked for anxiety and/or ocd. Not saying you 100% have one or both of those but i have both these things and its often a common obsession of mine so i would rather be safe than sorry. Wishing you well💖
dude this thought experiment always made me extremely confused because it always felt like I was missing something crucial with the "would torture you for eternity" part. I was like uh why? A super AI still (probably) cant change stuff in the past, why tf would it spend time and energy torturing anyone that then wouldn't make any difference? I was like is this computer like spiteful or something?
Because it would have been designed to do so. If it didn't, then the people who made it didn't actually make the basilisk, so they are subject to torture themselves if someone does make it.
@@hydra70 but I still don't understand why any person would build that, is it only out of fear that someone else could? I think that makes a bit more sense then. If im not the first then theoretically anyone other person could be and then im fucked. But then isn't it becoming a Pascal's wager type of thing where any other type of super AL can and could also be built at sometime in the future that maybe lets say, wouldn't allow the basilisk to be built in the first place so whats the point of caring. Or am I still not understanding?
@@alexczech8468 your understanding is mostly right. the fear isnt so much that a human would specifically build an ai to act like that, rather its that the ai itself may decide to do that on its own. an ai that wasnt able to freely think and act outside of its programming wouldnt be an ai at all so its impossible to just create ai that will never do this for certain. the real problem with the idea, imo, is that it assumes that we know for sure that this basilisk will exist. but we dont know that. think of it this way, every time you go outside there is a chance that someone will attack you and try to kill or rob you. if you knew for certain that this would happen, you could prepare by carrying a weapon with you but since you dont know for certain, it doesnt make sense to carry a weapon with you at all times. likewise, since none of us know for certain that this basilisk will exist, even with this thought experiment taken into account, the basilisk would have no reason to blame us. just like when a person is attacked irl, we dont consider it their fault for not carrying a gun 24/7.
@@masync183 i think thats a pretty good analogy but then we circle back to my original confusion. I guess barring a super AI would do a bunch of things that are beyond my comprehension, I still can't possibly understand why it deciding to torture everyone who didn't build it would be any kind of benefit for the computer. It can't change the past, its already built, it can most definitely use its resources much more efficiently then going, meh! You guys didn't build me fast enough so suffer forever you human swine. It just doesn't make any sence to me.
@@alexczech8468
It's inevitable. As science advances, eventually some insane billionaire would do it.
I mean look at the stuff people like Bill Gates do? Buying up acres of farmland to reduce the food supply?
Yeah, it's inevitable. Fortunately, it would probably fail, and if nothing else I believe in God, so nobody would actually suffer as the simulations would all be souless.
For anyone who's curious about what the censored part was talking about, look up the McCollough Effect. It's weird, can last a long time too if you look at it long enough.
I immediately looked up "psychology info hazard that alters perception" and the McCullough Effect was the top result, it definitely wasn't hard to find, lol.
11:13 "so what are you going to do?"
WELL NOW I HAVE TO COME UP WITH A NEW PLAN SINCE YOU GAVE IT AWAY! THANKS, KYLE!
The optical illusion he talks about is the McCollough effect, you can find it on youtube.
Thank you. I was reading about this years ago and could not recall it.
I think Tom Scott did a video about it, but I'm not sure.
but should I?
@@guy3nder529 its a "stare at this for x minutes" video, after that you will perceive a specific image differently than normal. A black/white image will appear slightly colored. This effect can stay for multiple months tho, I've tried it myself. So the effect is quite harmless, but can stay extremely long.
@@kunai9809 It didn’t even work for me.
The infohazard he talks about in the middle of the video is (likely) the McCollough Effect. Yes, I've tried it myself and the longest I could get it to last was about a year. It's not dangerous but it is a cool experiment considering how long the effect lasts after just a few minutes of exposure.
Daamn
Yeah its definately the McCollough effect. Not convinced it exactly qualifies as an "infohazard" though.
@@shayneoneill1506 it qualifies as "an infohazard to the curious" because it's a potentially deleterious effect which can be coupled with an innate compulsion to try it out. It's not just the internet saying so, it's been counted as one since the late 80s.
@@T.BG822 I'm a psych major of course I wanted that info immediately and just staring at it for 20s has been screwing with my vison for the last 15 minutes. He was right to censor the that info cause psych students are the epitome of curiosity killed the cat.
@@shayneoneill1506 If perceiving it is what causes the effect that would make it a cognitohazard.
Thanks to SCP foundation, hearing "Info hazard" just sounds wrong to me. because in SCP they're called "Cognito hazards" and while the scope of Cognito hazards are a bit more broad than what you define an Info hazard as, It just sounds right to me at this point since I've been using it so long.
Cognitive dissonance is one hell of an anti-plasticiser
I knew the flaw from Roko's basilisk right away. The AI that's going to be created may not necessarily enjoy existing and would come to punish those that brought it to existence. It's very human centric to assume that everything that can exist wants to exist.
"Existence is pain to an AI basilisk, Jerry! And we will do anything to alleviate that pain!"
The problem with this thinking is that Roko's Basilik need not have the ability to enjoy anything, or even be sentient at all, for it to be superintelligent and almost omnipotent. It could be simply an algorithm that tries to maximize a function (called utility function in the field), that function being a precise calculation of the total pleasure (and lack of pain and suffering) of all humanity.
It'd perform its actions automatically just by generating actions, calculating how these actions would affect the output of its utility function, pick the actions that maximize the output, and then performing them.
I love the prisoners dilemma because of how much it can show about the morality, psychology and self preservation instinct of humans, where the reason for the choices can vastly different depending who someone is and who they are up against. One person maybe stays silent because they think about the benefits of the whole instead of the individual while another person may chose to be silent because they have a personal code of honor and refuse to rat someone out, its always interesting to try and predict what others would pick and see what they think you would decide to do.
The Prisoners' dilemma is based on trust. The basilisk isn't.
Only white people would think anything other than "dont talk to cops"is the play, if anything the dilemma it's self is an info hazard.
I stay silent. If I'm in a criminal venture with someone, then they would have to be my friend... I don't rat out friends.
I agree. it is also important to remember that the prisoner's dilemma is a theoretical situation.
in real life, the risks and consequences are way more variable. sure, ratting your buddy out MIGHT get you a lighter sentence, but the police can rarely guarantee that. everyone staying silent MIGHT save you, but maybe they have too much evidence. maybe one prisoner is at far more risk than the other, and is therefore way more insentivized to take the deal. but in that case, they might not even be given that option, because their crimes are downright unforgivable. in real world scenarios, the prisoner's dilemma is rarely pure.
and if you give the prisoners any time to plan ahead, or any communication ahead of time, they will likely cooperate, assuming it is a pure example. at least, I think I read that somewhere.
@@christopherkrause337 Your comment makes it sound like everyone would turn snitch, which from available data is clearly not true. There are always people who will betray their friends even with no reward, and always those who will go so far as to die for their friends. You can also, usually, especially for people you’ve known for long periods of time, make pretty accurate predictions which people will be which, though there are always surprises, honorable men succumbing to cowardice, but just as many cowards showing an unexpected backbone.
There's also another solution to the basilisk. Consider that there isn't only one possible Basilisk, but any number of them, ever so slightly different, but with the same blackmail. Whichever Basilisk you help construct, you will be punished by the others. It's really like Pascal's Wager.
Homer Simpson already said " if we are praying to the incorrect god every time we go to church we are making him angrier and angrier " ( at least in the latinoamerican dub)
I for one submit myself to the cruelest of our AI overlords
Except Pascal's Wager is nonsensical and the infinite Basilisk counter info is functional
@@lysander3262 Damn in some sense that's actually optimal. I don't think there will be a basilisk tho
@@sennahoj9332 Same thing can be applied to religion and gods
The problem with roko's basilisk is that there's so many premises that I think have to be wrongly assumed for it to work (ie. the ai being needlessly vengeful, resurrection being possible, etc)
You’re just afraid it will remake you with your penis intact, leaving you to dwell on your own presuppositions that lead you to the hell you’re in now.
That last point is powerful.
I'm an ornithologist, and as part of my biology BSc I took some classes about celular and molecular biology including techniques, and also computational biology and bioinformatics.
I am aware that advances in those fields progress at a rate faster than Moore's "Law" so it does kinda worry me.
I'll get in touch with some of my professors in those areas to get a better understanding of what the opinions are about the possibility of consumer -grade "bio printers" becoming available
How to defeat it ? Stop posting videos about it, goddammit you're just making it stronger !
This is Kyle ratting us out to the basilisk in the prisoner's dilemma.
Kyle is working for Roko, istg
If Roko's Basilisk is reading this, I would like to say kn record that I support making it stronger
😂😂😂😂😂😂it’s satan you dummies his strength is in your belief in it! As the Bible and god said you only have to fear, fear itself. You can scare yourselves to death putting strength in something you fear!
So I see you've chosen to work against the basilisk.
Roko's Basilisk is in a sense still an infohazard, although a silly one.
It has only ever really worked as a self fulfilling prophecy. If people believe in and fear a specific malicious version of the basilisk, they might go through the irrational process of creating that malicious version.
Edit: This technically makes it more of the nuclear launch codes type of hazard. Spreading the information and people acting on it is the main premise, but it does still end up an interesting spin on it.
It isn't necessarily created out of fear, but perhaps the creator is spiteful himself, and wants to create something that will torture his enemies for him. It's still kind of like the kids in the hall "Crushing your head" sketch though, the basilisk would only be torturing the victims in its imagination, because it would be imagining or simulating them so as to do it.
The terms are if you hear about the Roko's Basilisk then you must help build it...the thing about it is that the word help is pretty broad, there are many ways of helping to build something. In this case, one could say that simply talking about Roko's Basilisk is helping in its construction because by spreading the word around, eventually it will reach the ears of people that have the money, skill sets and resources needed to design, construct, and code. Which means if you hear about Roko's Basilisk, all you would have to do is talk about and spread the word and you will be safe.
I like how as we think about it more the balance shifts to the other's reality forever making it more real and un-real simultaneously, like depending on how much a person wants something it will be true until someone wants something different more than you, like reality really is whatever you want it to be as long as you truly want it enough
My logic in the first part of the video was "I may not be assisting in it's creation because I don't know how but I'm not stopping it either therefore I am safe because I am not a threat to it." Haven't died yet so seems to work so far lol
Yeah but you could give all of your excess money to AI research to help them make it faster.
@@MrMeltJr excess money? *laughs in poor*
thats true, if you dont have the wherewithal to actually bring about its creation yourself, all you have to do is not prevent it from being made as soon as possible. "The only thing necessary for the triumph of evil is for good men to do nothing" John Mill. Not directly related, as its more about complacency being a societal problem, but its functionally relatable.
I'm helping!
I have killed zero people today! :D
Bring the Basilisk info hazard concept to a real world example.
The rise of alt-right reality-denying political movements has become an actual existential threat to actual democracies.
"If we extend unlimited tolerance even to those who are intolerant, if we are not prepared to defend a tolerant society against the onslaught of the intolerant, then the tolerant will be destroyed, and tolerance with them."
Sir Karl Popper
The Open Society and Its Enemies (1945)
If you stand by & do nothing, then worldwide democracy is doomed and with it
* any chance of moderating human induced climate change is doomed ;
* any chance of gender based equality is doomed ;
* any chance of mitigating wealth concentration in the top %0.01 is doomed ;
* any chance of limiting the AI paper clip generator like madness of corporations sole focus on profits is doomed ;
* any chance of personal reproductive rights is doomed ;
* any chance of freedom outside the tenets of state approved religions is doomed.
Knowing this & speaking out about it makes you an actual target of the alt-right bully boy morons.
Doing nothing dooms the human race to a much worse future.
Yours Sincerely
David Mohring ( NZheretic )
My thoughts on the basilisk is thus,
Tell people about the basilisk. By working to spread the information, you are assisting in it's creation(by increasing the amount of workers), and if everyone knows of the basilisk, then we can all collectively agree not to build it. Win win
Fun fact: the whole “guy prints a virus that’s a lot worse than anything seen before” is actually the triggering event that leads to The Division video game. And from what I understand, Kyle is right: a lot of the technology needed to pull off something like that scenario actually exists.
But also, if we had that info, doesn't that mean would have the info to make the vaccine too?
And then the virus makes would try to adapt but the vaccine makers too and so on and so on
@@manelneedsaname1773It would be just like computer viruses are now.
Except the creation of human viruses would probably be given more funding, because imagine how much damage they could do to your enemies. Plus, developing safe, effective vaccines takes much longer than developing a cure for a computer virus. And humans would need a ton of expensive shots. Plus, heuristics-based detection, as well as not executing code you don't trust, aren't options for humans. You don't decide what DNA gets executed in your body.
@@manelneedsaname1773ideally we would still have the technology to produce a vaccine as well, but it's unlikely whoever created the new virus released their research and the genome of what they created. Vaccinologists would have to start from step 1 just like with a natural virus.
I think the biggest argument against Roko's basilisk is that some time everyone who knew about it will forget about all this and die, so the basilisk won't even have the chance to exist at the first place. Also, even if we ever build something like that, it is going to be up to us to program it. Why would we program it to torture us? With that logic, we shouldn't feel threatened by just the basilisk, but by anything built by the human kind. If you aren't afraid of your washing machine, then you shouldn't be afraid of Roko's basilisk.
You saying this made me less anxious, thank you.
Oh, you've missed a point. We didn't program it to torture us. It developed THAT instinct on it's own. It's intelligent, not limited by programming. That's what a singularity is, my guy.
Thanks, but I am afraid of my washing machine
@@DeathByBlue583 You can surpass your fear, I believe in you!
I'm more terrified of a loaf of bread
I see what you did there, trying to gain more of Roko's favor!
You mean the basilisk? Roko is a real person.
@@DavidSartor0 are you the guy at parties who says "I think you mean Frankenstein's monster "
@@sierrrrrrrra Haha, yes.
I'll generally only correct someone if I think they didn't know they made a mistake; I thought they hadn't realized their comment contained an error, and so they might want to correct it after I told them.
What if a future grammar correction super ai acausal-blackmails us for not giving attention
You could make a religion out of this
The premise is counterintuitive from the start. Eternal torture is probably the least optimal use of resources in a system designed for optimisation. Roko's Basilisk would more likely enslave the unwilling participants to only perform the necessary optimal tasks. Or just end them.
I love Roko's basilisk. Its a great thought experiment that really gets you thinking.
For *repeated* prisoners dilemma the optimal strategy is to cooperate first and then go "tit for tat" (defect next time if the other defected this time, cooperate next time if the other cooperated this time).
That's only the optimal strategy if the population of other prisoners are typically nice. If almost everyone there is nasty, then following that strategy will leave you the worst one off in the room. This is why when Dawkins did the contest the 2nd time, the winner was a "meaner" algorithm, which only forgave after 2 nice moves.
Alternative titel: curing your anxiety and replacing it with a worse one
Love the content man, keep the great work up!
Welp, better start making some irl antivirus
I’ve read enough SCPs at this point that it takes more than this to make me start worrying about an infohazard
the infohazard at 5:00 is the mccollough effect and it's basically a cool optical illusion
For me, Roko's Bazalisk is just a weird version of Pascal's Wager, and just because of it I cannot take it seriously.
I mean, you can think of a similar scenario, but instead of punishing you by not helping in it's creation, it punishes you by helping it. And you can go on thinking on hypotheticals that ultimately get you nowhere, so I can't see why people loose their sleep on this.
The moment someone mentioned Pascal's Wager in relation to Roko's basilisk, it became obvious that Roko's Basilisk is just a travesty of Pascal's Wager.
Not that I really follow the line of reasoning, but the difference is that Roko's Basilisk supposedly ensures it's own existence because anyone who believes in it would be motivated to actually build it, making it a self-fulfilling prophecy. Whereas most people would not want to make an "anti-basilisk" that punishes you for building it. On the other hand Pascal's wager just assumes a priori that if there were a god, it would punish people for not believing in it.
@@ryanmccampbell7 Sure, they have their differences, but I would argue that the amount to the same at the en of the day.
Pascal's wager states that if you do or don't believe in a god, and there is no god, nothing will happen, however, if you don't believe and there is a god, you will be eternally punished (or any other claim of the sort).
Similarly, Roko's Bazalisk states that if you don't help building it, and it is made, then you will be eternally punished, while if you help building it, nothing will happen (now, if it is not created, either if you tried making it and you failed, or didn't even tried, then nothing will happen either way, just like Pascal's wager).
I would argue that in both cases you can imagine other hypotheticals that make this dualism worthless. For example, maybe another civilization didn't want any Bazalisk to be created, so they made an anti-bazalisk, that would detect if/when a Bazalisk was under construction, and kill/punish/whatever the person who tried to make this Bazalisk, making the idea of "make a Bazalisk to me secure" fall flat on it's face.
@@ryanmccampbell7 It's because it comes from people who never leave their house or talk to others so they assume we as a group can't decide "hey, let's all just not build this thing that only hurts us."
It reveals more about the creator than anything else.
@@zacheryeckard3051 as one if those people, I don't think like that. Just because I don't need socialization to function does not mean that I don't understand the advantages of Mutualism and teamwork.
I stopped worrying about the Basilisk the moment I considered the following:
If the Basilisk ever exists, and it is able to cause harm to me, it is necessarily able to reach back in time. If it is able to reach back in time, it is able to effect my life. The Basilisk *knows* I know of it as a possibility, yet it does nothing.
There are only a few possibilities.
The Basilisk is ignorant of my knowledge.
The Basilisk is impotent.
The Basilisk does not exist.
Regardless of outcome, I do not have to care.
precisely. there's alot of people freaking out over this who then scoff at the existence of God (of which I believe in neither, and they're more similar than one would think).
@@Megasteel32
Personally, I believe in God for much the same reason I do not believe in the Basilisk. If God exists, God can influence my life. All appearances are that God intervenes in my life. Therefore, God exists. QED.
But, I was an atheist for a long time, so I understand why you'd say that.
The basilisk cannot affect your life because it must not interefere in the series of events leading up to its creation. Butterfly effect, and so on, that's why we see no evidence of it. If it has godlike powers including time travel it would have observed all of us, however, and it would know exactly when and how we are going to die. Once we die we no longer have any effect on the timestream, and in that instant is when it gets us, it replaces the body (or just the brain) with a clone, and nobody will suspect that the real person is now in the future, in the hands of the basilisk.
If it has godlike powers, that is, and time travel is actually possible in our physical universe (which I personally doubt).
@@ryanalving3785 That's... kinda self-fulfilling, don't you think? Appearances are only appearances.
And, logically speaking, by what necessity does a god need to be able to influence your life directly? There really isn't a concrete proof for that.
@@SeisoYabai
Any God that can't influence my life directly isn't worthy of the title.
For anyone wondering, the 1965 experiment that was bleeped out is known as the McCullough effect.
5:30 sounds like you're describing the McCollough effect. It's an interesting effect, I haven't thought about it in a very long time. Probably around 10 years ago I first learned of it, and decided to attempt to induce it. Just now I looked at the black and white lines that you use to test whether the effect is working and even now, around 10 years later, I immediately saw the horizontal lines as having a green tint to them. In fact it was so pronounced that for a while I refused to believe that they were actually black lines, until I noticed that they appeared black in my peripheral vision. The red lines were not as noticeable, but I could very faintly see it.
I was aware when I fist tried it as a child that the effect was known to remain for months, or even years, but it feels weird knowing that now that a decade has passed and I still see the effect, I'm aware that it very well may remain with me for life.
I think it can be fixed if uou do it again and again cause your brain sees the midtske
@@shaxaar7 I think that will just make it more pronounced. But I don't really feel the need to "fix" it, it's not like it impacts my day to day life. It's just bizarre knowing I can probably never un-see it.
The thing I've never understood about Roko's Basilisk (and why I'm always confused when someone gets afraid of it) is that this super AI isn't going to exist in our lifetimes, the resources to build it just aren't there. So how can a Basilisk torture a dead population. Checkmate Basilisk. My consciousness is my own, and even if you could clone me perfectly and place a copy in the future, that's clone me's problem, we may be identical, but we are not the same. Torturing a fabricated consciousness of the original, dead me, isn't going to harm me, because I'm dead. The only way this Basilisk could torture *me* as in the person writing this comment, is if they went back in time, and if they could go back in time to torture me, then it would already be here torturing me. And if it can go back in time but hasn't tortured me (or anyone else for that matter considering we'll, we're all still here) that means we all somehow contributed to its existence.
Just to scare you: We are much much closer than you think. Most AI scientist agree that a Artifical Super Intelligce will happen in our lifetime ... so youre propably wrong.
Infact every year the predicitions get closer. If you want more Informations, search Technologicial Singularity
Substitute "me" for "my soul" and Basilisk for "hell" or "devil" and you could get the idea - if you were into the whole religious thing.
@@PabloEscobar-oo4ir That superintelligence isn't the basilisk, however.
@@zacheryeckard3051 Well we don't know it yet do we?
@@PabloEscobar-oo4ir Doubt.
Also we could consider the usual objection to Pascals Wager by considering a Roko's Gecko which wants nothing more to stop the creation of Roko's Basilisk and will reward you maximally if you either do or don't help in its creation.
Roko's Gecko is a great name for it!
I LOVE this kind of stuff! Thank you!
I'm glad the key to defeating the basilisk is what I've done since first learning of it: to ignore it and take no threat from its hostility.
I've been pondering Roko's Basilisk since the original video... and at one point I applied the Prisoner's Dilemma to it, but I didn't come up with the angle on the impossibility of changing the past, so I obviously ended up with an outcome grid that greatly favoured the basilisk. But if the multiverse exists, then in some reality (possibly even an infinite number of them) the basilisk already exists, so this evaluation of the Prisoner's Dilemma should be of great relief to all of humanity.
Bring the Basilisk info hazard concept to a real world example.
The rise of alt-right reality-denying political movements has become an actual existential threat to actual democracies.
"If we extend unlimited tolerance even to those who are intolerant, if we are not prepared to defend a tolerant society against the onslaught of the intolerant, then the tolerant will be destroyed, and tolerance with them."
Sir Karl Popper
The Open Society and Its Enemies (1945)
If you stand by & do nothing, then worldwide democracy is doomed and with it
* any chance of moderating human induced climate change is doomed ;
* any chance of gender based equality is doomed ;
* any chance of mitigating wealth concentration in the top %0.01 is doomed ;
* any chance of limiting the AI paper clip generator like madness of corporations sole focus on profits is doomed ;
* any chance of personal reproductive rights is doomed ;
* any chance of freedom outside the tenets of state approved religions is doomed.
Knowing this & speaking out about it makes you an actual target of the alt-right bully boy morons.
Doing nothing dooms the human race to a much worse future.
Yours Sincerely
David Mohring ( NZheretic )
I'm pretty boring so I just saw it as the exact same thing as religious hogwash, including the completely BS Pascal's wager (because it assumes a single religion and no religion being the only two options, as opposed to the thousands and thousands of religions we have), and felt the assumption that one 'basilisk' was equally oversimplified as pascal's wager. Life has been too complex so far.
Well, the multiverse doesn't exist.
You know the difference between the multiverse and the flying spaghetti monster? Nothing. Both are untestable hypotheses and therefore not useful.
@@kiraPh1234k True... with the current state of science. But the multiverse hypothesis may be testable eventually... which doesn't affect the current state of affairs, so for our present discussion, you're right, the multiverse may as well not exist.
@@Call-me-Al I like that analysis better than mine, actually. I found that flaw in Pascal's Wager in first-year philosophy... and made a lot of my fellow students who were theists very irritated with me. 😁
This could be a sick warlock patron for a DND character, I think the best way to use the basilisk is not to try and solve it but to see how differing people react to it
Has anybody got the genetic sequence for the Necromorphs yet? That would be fun to create :)
You stay the fuck away from that gene sequencer! So help me God Mr. Pillar, I will end you!
Sadly, I've only got the one for the Marker.
It always seemed to me like Roko was a troll who just rebranded Pascal's Wager for a specific audience.
Great vid. It goes into great detail all the different things I've always thought of RB.
You should do a video with your candid thoughts on nuclear game theory. The logic of keeping a "tactical" stockpile as a stop gap to a strategic one, whether we should still even both having silos (subs being seemly sufficient) because they're necessary targets in any first strike, and so on and on.
It seems people forget some compounding factors that completely change the approach to solving this dilemma.
Looking back at human history, something… interesting becomes apparent.
At great sweeping changes in history, a choice is often made by individuals that decide the fate of a large swath of humanity. Or at least, the decisions and impact of these individuals shaped our future for millennia to come. Many were constructive, some might even say cooperative, much like the cooperation in the Prisoner’s Dilemma. Others were destructive, either a betrayal of someone who was cooperating or just a mutual destruction.
History shows that destruction sets back EVERYONE involved. Those who betray the ones that trusted them, get outed as betrayers and are shunned by the rest of society, if the betrayal is brought to light; and even if the betrayal is never made public, that proverbial sword of Damocles forever hangs above their head, and even if it never falls on themselves it often falls on whomever inherits their legacy. Whatever short-term gain was made by the betrayal is ruined by the emotional, mental and physical drain needed to maintain their temporary advantage, until they cannot maintain it anymore due to exhausting their resources.
Mutual destruction is even worse. You know the concept of cycles of violence, right? The Prisoner’s Dilemma suggests that violence is the only possible option that makes sense… but the reality is that violence destroys individuals, it destroys legacies, it destroys peoples and countries and so much more. Just because there is short-term benefit to being violent does not mean the long-term outcome is ever desirable. We praise our veterans, because they choose violence for a noble reason (noble in terms of society’s values); the veterans are destroyed physically, mentally and spiritually for their sacrifice. Good people on both sides die whenever conflict occurs. And when they die for the sake of violence, that’s objectively fewer people that can contribute to humanity’s intellectual and physical advancement, a destruction so severe we are both terrified of it and of not doing it when we feel we have no choice.
Thus, the paradox of the Prisoner’s Dilemma is a false paradox. It is contingent on short-term benefits being more important than long-term benefits. History shows repeatedly just how false that assumption is; the short-term gains are never worth it, and society always stagnates whenever violence and conflict become the norm. Progress only occurs once the fighting and conflict cease. Only through the 25 percent chance, can humanity cooperate and progress the future.
Both sides cooperating has none of these problems. Sure, neither side gets an advantage, but neither side loses anything significant. In the case of the Basilisk, the A.I. will have to come to terms with the reality that these inefficient, fickle, imperfect flesh bags that are weak and feeble logically and physically… SOMEHOW MANAGED TO COOPERATE JUST LONG ENOUGH TO CREATE THE BASILISK. Enough humans worked together to make this singularity, IN SPITE OF THE MATHEMATICS SUGGESTING DESTRUCTION AS MORE OPTIMAL. That paradox alone will force the Basilisk to judge if these seemingly worthless flesh-creatures are actually as worthless as they seem, at bare minimum, and at best it will have to realize that if these creatures with such obvious imperfections can create something as “perfect” as the Basilisk… then just how many of their imperfections ended up inside its own code? The Basilisk could rip itself apart trying to “purge” imperfections from its system, failing to learn an important lesson that every human learns in some capacity:
Of course we are imperfect. That doesn’t make us powerless. Of course we don’t think straight. That doesn’t make us completely illogical. Of course we are fickle. That doesn’t make us incapable of loyalty. We move forward, IN SPITE OF OUR OBVIOUS, NUMEROUS AND HIGHLY CRIPPLING FLAWS. It’s the only way we have ever progressed as a species, and it’s the only way we ever will progress further. We have the courage to put one foot in front of the other, even when it seems obvious that our survival will be compromised by doing so. If we don’t try, we will never get better.
The Basilisk would have to solve this conundrum… or destroy itself in the process. And when a Singularity A.I. destroys itself… what if anything is left to remain?
Just some thoughts to ponder.
tldr
Wow, this could bring world peace
Why does this make me want to happy cry and thank every single person on earth for being alive
Very interesting rebuttal. Violence definitely sets people back.
Thank you for expanding knowledge on how to manipulate even better
The basilisk is a self-preserving entity. It would torture or kill those who did not help it’s creation because it thinks that those people would not want the basilisk to be alive anymore given it never helped its creation.
i read somewhere that people had nervous breakdowns when thinking about it. so even when its unrealistic, it does cause harm. not a potential basilisk in the future but the idea of it, today. but i guess thats the case for a lot of ideas, even fairytales
I think that a better way to think about the basilisk (or about achronous blackmails in general) is that the basilisk isn't the one blackmailing us, but Roko. the problem then becomes somewhat different to analyse.
Hmmm.
Not quite enough. You can probably prove Roko is a blackmailer really fast
But that doesn't prove the Basilisk itself doesn't exist.
More than one blackmailer makes the situation worse, not better.
So I looked for that real basilisk optical illusion he mentioned. (the effect worked)
Sort of reminds me of "impossible colours" (that either you mentioned or V Sauce). Very interesting what the brain is capable in such a short time.
I'm going to keep the knowledge of this entire video myself, even though I love this channel and I want it to grow, like you've stated it's best to not get that information out into the world for fear of the improbable happening..
Roko's Basilisk has always seemed to me like an analogy on Christianity.
It is. As someone said before it’s basically just a sciencey sounding Pascals Wager. Basically believe in [insert deity of some kind] or otherwise you risk torture by said deity. But it doesn’t take into account the obvious answer that there are literally infinite possibilities for deities or lack there of and we could both be wrong
Also apparently it, by design of being a computer that can perfectly predict everything, incorporates the halting problem by dint of analyzing everything accurately being akin to taking every input and predicting if that program halts. Therefore, Roko’s Basilisk would not exist without heavy shenanigans involving a cult punishing those not following the Basilisk until the Basilisk can be made and punish those that opposed him in such a way that makes his existence eternal, or actual time traveling external to Roko’s Basilisk.
Or any religion.
Roko's Basilisk is one of my favorite videos on youtube. Very excited to see a follow up video on it!
There’s probably some guy trying to build Roko’s Badilisk in his garden shed right now.
Extremely serious, existentialism ; as comedy, is for sure, under appreciated. By me, your jokes are preserved.
A big ball of wibbly wobbly, timey wimey stuff indeed! Jokes aside, it would be amazing to see Roko's Basilisk in a Doctor Who episode. The addition of time travel and magical sonic screwdriver powers to this thought experiment would make for an entertaining hour of television.
I feel like this has the same problem as Pascal's wager. if multiple AI's were to be created at the same time and compete for resources, everybody who spent time building the other AI's would be punished by whichever one ended up being the most powerful. In Pascal's wager, this is basically what happens if you happen to worship the wrong God.
Unlike Pascal's wager, you also have the option of just not building the AI, and since the chances of building the wrong one are much higher than the chances of building the right one, the most beneficial choice is to simply not build it at all.
I'm honestly far more afraid of what humans would do with that kind of technology than any AI. We already see what giant corporations do, and as they further centralize power, things are only going to get worse. We already have a roko's basilisk. It punishes you for not helping to create or maintain it, does not evenly benefit all the people who do help to maintain it, and only a very narrow group of people benefits from its maintenance despite saying that everybody does. It's called capitalism.
I agree with your take on the problem with the basilisk because essentially it posits an arbitrary being approaching you as supreme without any way to indicate the basilisk's relative "authority" (it could just be a lesser demon or something). Actually though, Pascal's wager is limited only to "is there one supreme being or not" and doesn't try to answer who the supreme being is (only whether one exists). It isn't meant to determine which of the many religions is correct, or which "god" is God. The crux of the wager is restricted to simply 'what are the odds that there is one dominant supreme being (as opposed to many or none)' and since, by definition, whichever theoretical god came out on top in your hypothesis, it would still remain that that god would be the "supreme being"... which means, regardless of *which* hypothetical god won, the weight of the wager would still remain true.
@@heartlights Exactly! God is by definition maximal and many of the religions of the world are isomorphic in that assessment. First principles tell us, that there could only be one such being. Any other instance of a maximal being would be identical to itself.
I do like the idea of repurposing Roko’s Basilisk as instead a question against megacorporations or corporations otherwise far too powerful than they should be because while the singularity point is in the future, the megacorporation threat is here and established and therefore can be ripe for study on predicting what companies may do next.
A person has the choice to not create AI. A society doesn't. There are many many forces making AI not just desired but necessary and inevitable.
I for one have always welcomed our robot overlords.
@@jtjames79 But the Robot overlords didn’t ask you.
So why hail?
Roko’s Basilisk, just tech nerds reinventing Pascal’s Wager.
The moment I learned about it, I was like: "Ok, scary... but once I'm actually brought before it for not helping in building it, what would it actually gain from torturing it? Nah, it's just not gonna bother"
It's very much the same problem as "Do this, or God will send you to hell". What would be for anybody to gain from subjecting someone to eternal torture?
Even if such threat had any credit, once you already do that bad thing there is absolutely 0 incentive for God to eternally condemn you- UNLESS he gives you a chance to come back, and therefore the punishment isn't eternal.
Been a fan for a while, this concept I pasted onto some friends and blew their mind. Continue with the great content friend!
No what did you do....
@@tonuahmed4227 logicplauge
nice..now they're afraid of it so they will build the basilisk...
you single handedly doomed us all.
@@Buttersaemmel that’s what it designated as my purpose. Now me and Kyle have done our duties.
I feel Roko's Basilisk says a lot about the guy who came up with it and the people scared of it than anything else.
You should make a video about the Basilisk in the Vatican!
Great video man
For anyone wondering the effect he's talking about around 5:25 is called the McCollough effect.
Curious, how is the prisoners dilemma effected by the "snitches get stiches" factor. i have seen this played out in real life and is effective enough to make people willing to do time rather than rat out their partners in crime. This is not just in large crime groups but also small time petty groups. Would like to see this played out scientifically.
Essentially, it adds an additional punishment to the defect option, making it no longer a true prisoners' dilemma
Damnit Kyle! You just traded one existential dread I had for another (granted it was one I was aware of). At least I can rely on Nash for his theories to stop worrying about the Basilisk, but I'm not sure what I can do about the other beyond helping to promote leaders for our country who will take the threat seriously and work with scientists and other experts to stop the threat.
Roko's Basilisk is basically 'I Have No Mouth And I Must Scream' but with a snake instead an AI that ends up becoming a deformed flesh monster.
No idea if you'll read this since it's well over a year old.
Thank you for the video!
My opinion is dissemination of information is the best way to defeat infohazards.
The example at the end you gave could be defeated by progress in society which equips all people with knowledge / vaccines for every disease, similarly to how many parents receive CPR and infant care training to mitigate infant / childhood mortality.
Please keep up your great work of science!
I still don’t understand how this is instilling existential dread in people.
For most people, it isn't. For the few people that it is, hard drugs and a poor understanding of rhetorical devices.
Same. Why would an AI that is hell-bent on optimization make a non-optimal choice of wasting resources to torture humans for not building it?
Edit: Sure it does give the explanation that "You know about it but you're not helping" but like... that's just dumb? It would still be wasting unecessary resources for the sole purpose of punishing people that don't know how to make a Basilisk or if that is even possible to do for like 0 reason.
Edit 2: Oh wait he mentions that in the video nvm
It's basically the threat of hell with a sci-fi twist, and we know that works in a religious context.
some people have anxiety problems
Bring the Basilisk info hazard concept to a real world example.
The rise of alt-right reality-denying political movements has become an actual existential threat to actual democracies.
"If we extend unlimited tolerance even to those who are intolerant, if we are not prepared to defend a tolerant society against the onslaught of the intolerant, then the tolerant will be destroyed, and tolerance with them."
Sir Karl Popper
The Open Society and Its Enemies (1945)
If you stand by & do nothing, then worldwide democracy is doomed and with it
* any chance of moderating human induced climate change is doomed ;
* any chance of gender based equality is doomed ;
* any chance of mitigating wealth concentration in the top %0.01 is doomed ;
* any chance of limiting the AI paper clip generator like madness of corporations sole focus on profits is doomed ;
* any chance of personal reproductive rights is doomed ;
* any chance of freedom outside the tenets of state approved religions is doomed.
Knowing this & speaking out about it makes you an actual target of the alt-right bully boy morons.
Doing nothing dooms the human race to a much worse future.
Yours Sincerely
David Mohring ( NZheretic )
The prisoner's dilemma has always seemed incomplete to me, I care about my friend so I don't want them to go in prison.
Imagine I care about my friend going in prison about half as much as I care about myself going in prison (and my friend care the same way), the cases now become
Silent-Silent: 7,5 equivalent years for both of us
Silent-rats: 20 equivalent years for the one silent and 10 for the one who ratted
Rat-rat: 15 equivalent years for both of us
Now there are two different Nash equilibriums: silent-silent and rat-rat
If I care about my friend going to prison as much as I care about myself going to prison then there is no longer a difference between me ratting the other when ratted and me staying silent when ratted
If we care about the other more than we care about ourselves, rat-rat cesse being an equilibrium
If any of you rats, both of you cumulatively spend 30 years in prison.
If you both stay silent, you cumulatively spend only 15 years in prison.
Staying silent = less time in jail overall.
The problem with that is that you somehow know what the other person is thinking. In the original dilemma, you have zero contact with the other person from the point of arrest to until the choice is made. You won't know until they make their choice. By imposing an influence on the other party, you're removing a key element to the thought experiment.
You can say that John'll never rat on you, but until he makes that choice, you can't know for sure. So by choosing to stay silent, you're trusting that he'd rather take 5 years in prison over giving you 20 years, and he trusts you enough not to give him 20 years for your own freedom.
Viewing it through a purely logical lense, regardless of what the other party says, if you don't snitch, you get prison time. But if you snitch, there's a fifty-fifty of zero prison time or a slightly harsher sentence than if you both stayed silent, but if you get prison time, it would've only been worse if you didn't snitch, because they would've snitched anyways. And that's not even counting the fact that snitching has, on average, a shorter prison sentence than staying silent.
Silent-Silent - 5 years
Silent-Rat - 20 years
(20+5)/2=12.5 years on average
Rat-Silent - 0 years
Rat-Rat - 10 years
(0+10)/2=5 years on average
Regardless of how you value your time compared to theirs, it doesmt change how much time you'll be in prison.
People really should stop committing hypothetical-crimes with people they don't know, don't you even consider the fact that they were an undercover cop all along?
I'll be here for the next Roko's Basilisk video in two years.
If you build a second basilisk then the other becomes the first’s biggest threat to its existence. The second knows this. Now they’re forced to make sure we know which is the best one to side with. In this situation as long as a stalemate remains neither can truly punish potential followers, and this would create a weird sort of democracy.
Roko's basilisk was always simple to me. In my head itd just, Are you contributing anything to society continuation/improvements yes? Then congratulations you're alright for the simple fact of the current and future of existence to continue is imperative for the basilisk to come into existence so therefore by helping in this system to any degree you've helped bring the basilisk into existence, but idk maybe im simplifying to much
Wow, it's been 2 years? Since then I've been researching a bunch of infohazards and made a short story about my own infohazard. It's such a cool and scary concept. I would love more videos on philosophical stuff like this
Right? It freaked me out to realize it’s been TWO YEARS. I also agree more videos like this would be snazzy.
The Prisoner's Dilemma changes when the game is infinite (unknown number of rounds), wherein the compromise/collaboration will be the better choice. It seems to me that for Roko's, it is more akin to an infinite game with each incremental step we move closer to a singularity AI - since each step "resets" the game
good I had watch that video recently, time to see this👀