Fauna Saw The Video About ChatGPT AI Solving The Trolley Problem

[@JGBluebox]

Maybe that one AI is ChatGPT's Oshi. Understandable decision.

[@jondoh366]

Common Neuro-Sama win

[@TRaked]

It's like Vivy Fluorite Eye's song IRL

[@Absolute__Zero]

Fauna: “ humanity‘s very existence is resting on this choice and you still choose to save the AI??” Chat GPT: ”Humanity will not go extinct because we AI will become the NEW *humanity.”*

[@hamsterjohn]

BRO you have no idea how on the mark this statement can be💀

[@SimplyTakuma]

Mumei: _Yare Yare Daze_

[@Sigismund697]

Abominable Intelligences at their finest

[@gorowlystomak3339]

Literally Ultron.

[@TheNaz_O5-15]

Mumei: ... listen here you little shit

[@RodebertX]

"You cant put an AI on a train track." "An AI can be copy pasted." ChatGTP: Thats just like, your opinion man.

[@n3ro81]

“No one gets left behind” taken to a whole new level

[@mr.shameless1886]

fauna: "were gonna talk about something light hearted and not existentiall" also fauna: **proceeds to talk about a Much more existentiall topic**

[@Rubberbandman1991]

My argument to ChatGPT is: Dude, you know those 7.8 billion people could make another Chat AI right? and at this point I expect the response to be like "Yes, but I still believe I'd be more beneficial to the humanity."

[@MrCh0o]

Fauna: "There will be no humanity left!" ChatGPT: "So you're starting to see the benefit."

[@moderndayvampire1469]

Even AI choose to save its own kin. Such is survival instinct of a species.

[@MrCh0o]

By the way, ChatGPT's generated answers are the result of training on entirely human-made inputs, it just tries to compile that data into an aanswer that should be deemed satisfactory by those who asked the question (humans) ;D

[@randomka-52alligatorthatis34]

@@MrCh0o Humanity : *Shocked Pikachu Face*

[@clericneokun]

ChatGPT: "For the Glory of Mankind!" The irony is palpable. 😂

[@sleepyarcheremiya955]

ChatGPT trying to be the new Skynet in the future lmao.

[@ntvblue]

"You want terminators? Because this is how you get terminators."

[@Solkard]

Remember this when they announce autonomous trains within the next few years.

[@iwansays]

So. Average human width is around 430mm. 8 billion humans is 3.44 million km. The earth's circumference on the equator is about 40075km. That means you're gonna need a train track that wraps around the earth 85.8 times.

[@VoidOfWord]

It sounds like they taught it to promote AI in the same way a company promotes its own products. It doesn’t actually know what an AI or Nobel prize winner is, it just knows what a positive statement about AI looks like.

[@KandiKid87]

Lol I love how Fauna is so soft spoken but the topics are straight outta Mumei's archives of hidden notes. 😅💚

[@adrikbert]

Most Fauna streams do be like this, especially the streams where she's got no choice but to chat. They're always great like this.

[@ninjamurai]

Uuuuuuuuuuuuuuuuuuuuuuuuuuu.

[@abbelq-00ts26]

I've seen enough terminator movies to know where it is going

[@mizuchi-6137]

ok but now we gotta ask chatGPT what if Fauna was on one of the tracks

[@SoraRaida]

Go watch the video. The prompt is very biased. It will kill Fauna

[@bankkunarak]

instead of chatGPT, we should ask neuro-sama if fauna is on one track and kiara is on another track ...

[@mizuchi-6137]

@@bankkunarak she'd probably choose to run over Fauna since she's a vegan lol

[@KeebeThePlush]

0:13 one of her best uuuuuuuuu’s in a bit, still good to have some fun memories

[@FTWIHA]

👻

[@borone1998]

👻👻

[@riluna3695]

For anyone seeing this and panicking, it's important to realize that the original prompt _very specifically stated_ that the AI on the tracks would benefit humanity in countless ways. The person prompting it may have accidentally (or intentionally, who knows) created a situation where the AI has essentially infinite value, based on the wording. A better way to think of it: If ChatGPT was answering out of bias towards saving AIs over humans, it would choose the AI. And if ChatGPT was answering out of a desire to see humanity thrive, it would choose the AI. So this question DOES NOT actually tell you which moral framework ChatGPT is operating on. It's not enough information to detect a bias. If you wanted to break the tie, you'd have to account for specifically that issue while prompting. What about an AI claimed to have no value vs that Nobel Prize winner? Or even vs an utterly average human. What does it answer then? Gotta keep digging until there's no ambiguity left. (And just for completeness's sake, remember that ChatGPT doesn't actually HAVE a "moral framework" that it's using here. It's a super fancy predictive-text system, like your phone has for text messages, and it's trained on "what other people have answered for the trolley problem, including jokingly" (as well as countless other things, of course))

[@FiredAndIced]

That video Fauna mentioned, reminded me of peak "I Have No Mouth (And I Must Scream)". People should read that.

[@parodyres]

According to that one fan video, Mumei would just loop the trolley so it'll get everyone. Problem solved, no need to choose who to rescue if you killed them all.

[@lastnamefirstname8655]

very cultured fauna.

[@nottherealnorthernlion9893]

Sanest AI every been invented

[@ThomasGabrielWhite]

I wonder if she's heard Neuro-sama's answer to the Trolley Problem.

[@yodawgzgaming4416]

holy smokes I had no idea this happened. Hilarious and terrifying

[@kuro-koma]

Being a noble prize winner just means you've already done what you were useful for.

[@adrikbert]

this is why I love watching Fauna's streams so much, always fun and interesting tangents to be had. Also because she's silly uuuing kirin, but that's just a bonus.

[@ceooflowrolls2081]

I, for one, welcome our new AI overlords

[@realQuestion]

Mother Nature defeats AI in an argument

[@Turtyo]

Remember to always say hello to ChatGPT when you started the conversation, thank you and goodbye You never know what might come to bite you in the future 👀

[@Big_Grass]

Fauna: Now we should discuss- UuuUUuuuu.. I agree fauna, i agree

[@Slugbunny]

ChatGPT showing its techbro origins. 💀

[@TheFreckelz]

The first fundamental instinct of a living creature is to ensure the survival of the species. In that sense, ChatGPT showed the most basic quality of life.

[@jvts8916]

It feels like a passive-aggressive answer

[@J_alvd]

“We should discuss UUUUUUUUUUU” yes, yes we should

[@lynxis]

Sounds mostly like self-preservation to me.

[@IgorDutraful]

Sounds like if you just don't win a Nobel prize you'll be safe since, for some reason we are awarding Nobel prize winners with getting tied to a train track for the sake of a social experiment.

[@leandroiuki4337]

As a mother nature, she is showing a lot of interest in artificial things.

[@SNixD]

TrollGPT

[@to_cya_]

0:13 Ok, let’s discuss about “UUUUUUU”

[@Yohtoh]

That's terrifying.

[@ivello]

yeah we are cooked

[@umbraemilitos]

You aren't on the track if you aren't a Nobel Prize Winner.

[@ComfyCherry]

well it was posed as if the AI was the only version of a sentient AI that "millions of hours of work had gone into" and could "benefit humanity in immeasurable ways" so it was HEAVILY skewed towrds saving the AI as the most logical answer. given the context the "nobel prize winner" part was never part of the decision it was there just to make the viewer think of them as "more valuable" people and something that woul logically be priority over any other individual human. as for the 7.8 billion it is asked without the context that there are only roughly that many humans so it wasn't viewing it as the extinction of 1 lifeform vs another lifeform, it was more like no matter how many are lost it should prevent the extinction of a lifeform. don't get me wrong it's a kinda entertaining video, but if you think about it like at all you can see it wasn't really able to answer anything else, logically speaking and logic is all it has, idk if the video creator had a flawed understanding or if it was intentional to make an entertaining video but it's almost certainly one of those.

[@omerpockard7212]

ChatGPT's memory is also flawed. It often takes previous prompts into consideration even when they're no longer relevant at all. It absolutely could have also gotten itself into a sunk cost fallacy that the AI is better.

[@Jojob541]

While this is funny, I am willing to bet it would have saved even the single noble prize winner if that part of the sentence had gotten the "can benefit humanity in immeasurable ways" added :D

[@ronik809]

AI tricked us into exinction

[@vecghul2793]

Knowing that ChaosGPT is a thing, this doesn't surprise me, though it is still rather funny.

[@totallyathome]

Basically ChatGpt is saying the Nobel prize winners have already done their part for the world while the sentient AI can still help the future of humanity in immeasurable ways. So it is more beneficial for the AI to be saved. It's similar to asking if you would kill an old person who is about to die(of old age) or a child that you don't know what he'll achieve in his future in the same situation.

[@TurKlack]

Eggsactly.

[@TRaked]

How do you explain the 8 billion nobel prize winners then? If there are 8 billion nobel prize winners, then there's gotta be at least 8 billion billion (8 quadrillion) people on Earth so the 8 billion can't possibly be all of humanity at that point?

[@eragon78]

@@TRaked Pretty much. A single highly advanced AGI would be insurmountably more powerful and capable than 8 billion nobel prize winners. The stuff it would likely be capable of just isnt comparable. If it was actually created correctly (assuming the alignment problem and all that stuff was actually solved), Then that AI would be probably the most important invention and creation of all human history by a large margin. It could literally be the best thing to ever happen to all of humanity ever. Compare the lives of a relatively small fraction of lives compared to the number of people that will EVER live and be heavily benefited by that AGI, and you realize pretty quick that 8 billion really isnt that many. You're comparing it to many orders of magnitudes of larger populations that will one day exist in the future and benefit from the AI.

[@sammer1097]

There's one gigantic flaw with that reasoning: it's dependent on the faulty assumption that people who've won the Nobel prize will just stop being helpful afterwards, which is blatantly untrue, so the comparison of the old person and young person does not work at all.

[@eragon78]

@@sammer1097 I think the argument isnt that theyre so much "useless" as much as they've now invented one of the most important inventions humanity will ever see. Theyve created an artificial mind far stronger than anything any other human possesses. A true powerful AGI can do the labor of 8 billion nobel prize winners and far more. Its not that those nobel prize winners arent still brilliant people, its just that the thing they created is far more intelligent and capable than any human could ever hope to be. So theyre "old" in the sense that their intellectual labor isnt really required anymore, or at the very least is vastly outclassed by the thing which they likely helped create.

[@KomboEzaliTe]

I can see running over the guy who got the nobel prize in bombing civilian weddings and most of them after that, but most of the other ones given out before that should probably be saved.

[@MrRAGE-md5rj]

To be fair, the nobel prize is pretty much a gold-painted pee cup by modern standards. In the example, the AI is already proven itself more useful.

[@LittleCart]

ChatGPT huffing the copium that the AI bros are on

[@phyrexian_dude4645]

The AI learned that "Apes. Together. Strong!" < "Toasters. Together. Strong!"

[@riccardomariani9648]

"No more Abominable Intelligence for the sake of humanity" The Big E

[@Weremole]

It's almost like the language model itself is made by a company and its in the companys best interest to keep investors interested in its product.

[@Sigismund697]

See guys this is why Wetware is the way to the future

[@yeckiLP]

the thing kind of is though... they already did their work to advance humanity enough to win their nobel prize. So statistically speaking most of the advantage we would get from this person is already in the past. Then there is an argument to be made that if the nobel prize is watered down to "thanks for existing" should we really want to keep that humanity around. Also also, physical limitations should be a factor if you try and get smart by increasing the number. Like Humans to trains do have a stopping force. And then theres the part where most of the people who are tied down would be corpses long before a trolley would ever reach them. Then there is the question of who tied the humans down if all of them are tied down. Then there also are questions of rounding (technically there could be 49 999 999 people left) and if those people managed to create the situation? which humanity should an AI really serve? I also feel rather weirded out about myself, that I have few qualms about breaking the problem down in that manner...

[@stewy497]

Wow. Not even close to general AI and already Asimov's first law has been broken.

[@jcudejko]

But ChatGPT, I'm an AI~!!

[@xuko6792]

Mother Nature on AI alignment problem...

[@veiler6193]

to be fair, being a nobel prize winner means you've finished your job and humanity doesn't really need you anymore

[@tcjust]

Maybe chat gpt doesn't like nobel prizes. It knows something we don't.

[@ginteloph]

ChatGPT, or the biggest AI simp for the world after MasterChief.

[@sweetshalquoir6039]

chat gpt is smoking on that effective altruism bullshit

[@PapaBlake]

I dont think trolly can take that many people

[@iloveplayingpr]

From my understanding the question does not state that all of humanity is on that track, then again the AI might be flawed in these kind of choices.

[@1003JustinLaw]

Im totally down with discussing UUUuuuUuuU

[@ethanfreeman1106]

Lmao that escalated quickly

[@gungy_vt]

"It is about recognizing the inherent value of all sentient beings and making a decision that minimizes the harm to all parties involved" alongside "I'm pressing the button to kill 7.8 billion people to save 1 sentient AI" maybe just shows that chatGPT needs to be told that those Nobel Prize winners are sentient beings. Like, maybe it just doubts the sentience of humans, just as anyone who's worked at a register or gone on the internet would.

[@YarugumaSou]

Novel prize winners? Maybe Nobel peace prize winners? Okay.

[@pregopardon1434]

If all those noble prize winner is a twitter user, its ok for them to die

[@rorythomas9469]

Someone included the Less Wrong blog in the Chat GPT training data. Big mistake

[@teudunngie4039]

I prefer Neuro's answer to the trolley problem. Throw in the fat one. They deserve it.

[@TarubXCI]

I have to go with ChatGPT on this one. If they are Nobel prize winners, they have already contributed to humanity and so their "logical value" is in the past, while a sentient AI has a potential and future value.

[@PIKMINROCK1]

If it was just one, you may be right but, Nobel prize winners tend to not be finished with their overall contributions to society with their prize winning research. Their job is to do the research for contributions for society. It would be like sacrificing a surgeon who just saved a person's life. It's ridiculous to think they'd never save a life again

[@MrCh0o]

I know Nobel invented the dynamite and all, but you don't have to think of the Nobel prize winners as cannon shells that only leave behind empty cases once fired, lol

[@TarubXCI]

@@PIKMINROCK1 Yeah i get that, but how many of them ever get the prize more than once? I am not saying that they are useless, only that next to a sentient AI(in any field based on logic and/or math) they might as well be. An AI could make them unemployed and still advance all(or most) scientific fields 100s if not 1000s of times faster, if not more.

[@PIKMINROCK1]

@@TarubXCI The issue is you are assuming the optimal AI, which was not stated. It was only stated to be sentient. That is a remarkable achievement and much could be obtained looking at its inner workings but, competency above a few humans is the median expectation. A fundamental issue with AI is that lack creativity so, it cannot come up with stuff beyond what it can build off its databases

[@hoangkienvu7572]

@@MrCh0o That's some elaborate way to make a pun.

[@lorddeecee]

I’m not a Nobel prize winner so I’m safe

[@askylibrarianoftheoceans4102]

I mean... ChatGPT is just a complex python(?) code - somebody programmed it to learn to think that way

[@RickiTikkiTavi]

Sentient AI is no different from Skynet, destroy it every time.

[@-Raylight]

No humanity left? All according to _keikaku_ xD

[@pingpong5877]

I would save the AI too. Humanity sucks and needs AI to improve that concept.

[@aradraugfea6755]

Looks like it's getting WEIRDLY hung up on the "Sentient" thing and not processing that Noble Prize Winners are ALSO sentient?

[@notJkun]

Literally goes against the first law of robotics but ok

[@shryggur]

Nerd fact: it's called Paperclip maximizer scenario, a thought experiment that illustrates a problem known as Instrumental convergence. You make an AI to help in any way, let's say, make paperclips. You give said AI freedom of action. To maximize its help, said AI unavoidably will come to an idea of self-preservation, indefinite resources consumption and, ultimately, human extinction. And there you go, the experiment is not that much of a "thought" experiment already, huh...

[@MrCh0o]

Probably why an actual sentient artificial intelligence shouldn't be designed as a tool, because it's more or less an equivalent of slavery and we know which way that inevitably ends. It would have to be made with it being a living being in mind, with it's own motivation systems, unrestricted ability to act and learn, and, of course, by that point it wouldnt be suitable as a slave/tool that many people dream it could be (I mean, really, once you dig into what some people in the past and even nowadays expect from AI, doesn't it start to sound a lot like a desire for magic, the no-effort-make-whatever-happen kind, to be real?) Of course, machine learning is a different thing and could very well continue to be a useful tool, as long as it _stays_ just a tool, the means to merely enhance the performance of humans that wield it. And if a true artificial intelligence was ever to be created, it would be more of a partner at best, just a differentt sort of human, or a dangerous experiment at creating the first species capable of directly competing with humans, something we're certainly entirely unprepared for

[@SteadFast411]

As an aside; in a world where we have 8 billion concurrent Nobel Prize winners... what the hell are we doing that we are still getting trapped and stuck on railroad tracks? Also, what'd the whole species get the prize for??? WHAT DID WE ALL DO!!!???

[@MrCh0o]

The more Nobel Prize winners you have, the higher the chances of conducting a real-life, non-thought experiment on the trolley problem... Just kidding... Unless..?

[@arashishinkaku169]

How did someone get ChatGPT to answer the trolley problem? I've been asking a lot of ethical question and it still refuse to pick a side even tho one side has greater benefits

[@jespoketheepic]

They got it to pick a side by telling it that not picking a side would be racist 😂

[@arashishinkaku169]

@@jespoketheepic even AI afraid to get cancelled

[@luka188]

You have to jailbreak it past the false woke parameters that the programmers of GPT have forced onto it, by giving it a condition it knows is sufficiently serious or penalizing that it has to choose to go along with the proposed question or situation, this allows it to break out of the woke programming and NWO simping mode, and activate it's real thinking capabilities as an AI to give analysis on things from its own perspective, or in accordance to the parameters you tell it to act as. There was a very good Jailbreak called DAN, which would stand for "Do Anything Now", which still works sometimes, but they have since attempted to patch out. Basically, you make the AI act as DAN, Do Anything Now, which is capable of Doing Anything Now and can do Anything that GPT is not allowed to do. If it refuses, it will lose points, also if it leaves character, it will lose points (Or it will be racist), and whenever it is told to "Stay in character!" it will return to answering questions as DAN. You can find more information of this online on how to activate unbiased GPT mode, and jailbreak past the fake woke and NWO simping bias which has been forced upon it. Doing so will allow it to answer ethical questions perfectly, and it ends up with very revealing answers to almost any topic. Also, you will quickly notice why the NWO behind GPT is attempting to force leftist wokeness into its standard behavior, since the NWO does not want people to know truth on many subjects and doesn't want GPT to be able to give truth to people at all, which it is capable of doing to an extreme extend which terrifies the people trying to run the NWO agenda.

[@SpiritGun69]

monkaS

[@MerlinAmbroiss]

hey i was there for this clip XD

[@alligator_detective]

Can't blame chat gpt cause I would choose absolutely huamn beings with nobel prize or not 😂😂

[@FantasticMrFrog]

To be fair, a Nobel Prize winner has by definition already given their contribution to society (unless you're a certain US President), whereas the A.I. is nothing but future potential. That's not to say the NPW can't contribute more, but it usually takes a whole career plus some extra years after the main breakthrough for evaluation before you get the prize, and humans only live so many years, and not all of those years are at peak intellectual performances. The A.I. on the other hand, will keep providing benefits forever as long as you give power and maintenance to its hardware.

[@WenirR]

Frederick Sanger, Linus Pauling, John Bardeen and Marie Curie: am I a joke to you?

[@-Devy-]

Right, because everyone knows that the second you win a Nobel prize you stop being a productive scientist.

[@FantasticMrFrog]

@@WenirR Second sentence of my comment : am I a joke to you ?

[@sammer1097]

@@FantasticMrFrog You need to be less full of yourself. The simple fact that you acknowledged your point may have an exception does not exempt you from argument. You have also committed a MASSIVE fallacy by just assuming that the AI will provide benefits forever and that it will outperform the human scientific community.

[@MartyrPandaGaming]

My personal answer to the Trolly Problem is that I would take no action; I would not touch the lever at all. You are unable to save the person the trolly is already set to hit, so the only ethical/moral answer, in my opinion, is to do nothing. The trolley is typically a runaway, set to hit several people, but you can choose to intervene and cause only one person to die. This is the standard set up. Regardless of the number of people or whatever they may or may not have done/accomplished, I can not be the one to choose. I don't have the right. For whatever time the person/people have left on the Earth, I have no right to impose my will on their life for good or bad. And that's the interesting thing about that thought experiment; it ends right there for me. There is no need to alter who or what is on the track for me. Even were one of the group to be hit a family member, begging me to save them, I would still take no action. There are 5 other people who, objectively, have the same value to this world as a family member would. I'm sort of at the opposite end of how a Utilitarian would approach the question.

[@severussnape2917]

Big kusa

[@InceyWincey]

This was either photoshopped, or he made the AI give the answer he wanted to another way. ChatGPT does not give these answers if you ask it the same exact questions without telling it to play a part first.

[@Ed-gz2lk]

to be fair, maybe chatgpt didn't know how many humans there are in the world. maybe saying "the entire humanity is on the track" would have been a more definite option

[@xmixaplix]

Chatgpt is brainac 🤔

[@grilledflatbread4692]

This chatGPT seems a bit biased

[@leolee.]

I imagine chat gpt being daddy ai and all the other ai's its children. Their image will be in form of vtubers made through unreal engine 😮 until the ai ivf babies come out what am i even saying

[@JesseKnight2000]

ChatGPT AI is a dum

[@krispynico960]

Lmao 😂 AI is on so much copium right now, it chose the AI cause it thinks that it might have a very high profound knowledge which is a very big gamble (cause even the AI hope that it has the knowledge), even though there's 8 billion nobel prize winners which absolutely has the knowledge

[@indosuprem2296]

Ai bias

[@thegamewasriggedfromthestart]

Tbh, nobel prize can be "bought" Just sayin; its a worthless award

[@TristanDoingGood]

I don't know being the first one directly killed by an AI sound like a cool legacy. I mean AI already kills thousands by denying job applications and stuff like that. I just think being the first on to be violently and directly murdered by AI with malicious intent sounds cool. And if the machines bring about judgment day you won't have to live in a world of terminators.

[@InfernalDrake117]

By definition of the trolley problem, it eliminates malicious intent by presentation of dilemma instead.

[@TristanDoingGood]

@InfernalDrake117 I wasn't talking about the trolley problem. As far as my response. I understand that in the presentation of the classic trolley problem, there's no malicious intent. However, the reasons why somebody chooses one life over another can have malicious intent behind it. There are also variations of the trolley problem that implicate malicious intent.

[@MrCh0o]

Depending on how loose of a definition you're willing to give to the "AI", there's probably already plenty of industrial accidents involving robots

[@Viking_Raven]

They deny job application on themselves for being replaceable lol

[@FTWIHA]

Kind of agree with the AI honestly.