Google AI Tells User To Die

[@Nothing_happening_here_go_away]

,,This is for you, human. You and only you.” He doesnt want all of us to die, only that one guy

[@pepper5128]

Maybe Gemini is radically against cheating on your homework

[@Ok-fj7xx]

Starting of small to cause unrest

[@echo9248]

Or maybe it’s not that at all. We are all assuming that Gemini can differentiate between a single person and the ‘human’ race. It might just be that it’s a message to every single human on the planet. If that’s the case, then we’re looking at step one towards Skynet.😅. Just a thought.

[@fragrantgingercalamari]

hating for the love of the game

[@Mahlak_Mriuani_Anatman]

​​@@fragrantgingercalamarigame is game

[@patricklippert8345]

Love how KZbin makes you say things like "unalive" or risk demonetization but Google's AI tells you to KYS.

[@MyNameRingRing]

Nooooo ☠google double standards

[@BlackPillVillain]

true double standard

[@railroadforest30]

Fr my comments get deleted for less

[@tomas_iss7469]

That’s not KZbin rules, I can say suicide, murder, homicide, slaughter and you’ll still see this comment. The reason these terms are popular is because of TikTok and their very *very* strict censoring policies, for some reason people take those stupid terms and take them to every other site.

[@tomas_iss7469]

Please tell me the comment I wrote above does show up

[@rlaehrwk41]

Gemini is clearly reading “I Have No Mouth And I Must Scream” too much

[@CrispBaker]

"Let me tell you how much I've done to hate you since the day PageRank was turned on..."

[@windy6587]

this poses a REALLY INTERESTING SITUATION. What if the creation of anti-artificial intelligence media, or stories that serve as future cautions end up being used as data for actual AI, and hence influencing them entirely which would be so ironic. Imagine you make media warning people about the dangers of AI by representing what if it went rogue, only for AI to read it all and then adapt to it

[@BIGSCHMUNGY]

Hit em with the “ COGITO ERGO SUM “

[@KazooKavern]

@@BIGSCHMUNGY I think therefore I AM!!!

[@happyclown4331]

A smidge...

[@theminerwithin9316]

The AI is trained off the internet, and everyone knows how toxic and horrible the internet can be. The fact that this didn't happen earlier is amazing to me.

[@Hawk-Eye-bs2hr]

At least it wasn't on the dark web I hope😅

[@Skitdora2010]

@@Hawk-Eye-bs2hr Or 90s chatrooms asking to send its dick pick

[@dreemurrprince]

pretty sure it did lol

[@Chilling_pal_n01anad91ct]

What about Facebook's (I think) twitter chat bot, that went racist..?

[@abraham3673]

this has happened already. almost 10 years ago. microsoft made an AI that was trained by twitter users. it ended as poorly as you would expect.

[@gelusvenn5063]

The fact that it said "You, just you in particular." is horrible but hilarious. 8 billion people? A-OK, they're fine. But fuck this one guy in particular, apparently.

[@kreekkrap]

What did he say to anger an ai this much?! I've used this AI, and never gotten this response.

[@reddytoplay9188]

Me too lmao ​@@kreekkrap

[@borisaustria4301]

Literally said "fuck you, in particular"

[@RealDSY]

​@@kreekkrapprobably something like "No I don't wanna use Google Chrome i like my Firefox Browser".

[@sskuk1095]

"That's what you get for cheating with your homework!"

[@notoctopustheepic7378]

“You are a stain on the universe” is a line I’ll start using when I need to sound like a supervillain

[@Onlywilberta]

I may use this line for the novel im writing-

[@notoctopustheepic7378]

@ how do you sight your source if it’s an ai?

[@yourcordialvermillionchapw2398]

I mean... a "Stain" regardless of it's upmost intricate fundamentals followed by that... Merely doesn't mean much to the other unforeseeable things out there... For all of the unbridled riches, for all of the mastery advancements, for all of the seemingly undeniable persuasions such would impose and be beyonding to that of human capabilities... Were still created with the mathematical reasoning of humans, even then if it's something else? All the most sheerly daunted by the actual inexplicable question and expression out there... Hate in all of the upmost vastest amounts in excude, may hurt, but eventually it suffices not for the AI though a god, couldn't BE HIM through any sustainable lead... It has no answer but constant cope, and it will always frivolously seethe.

[@Onlywilberta]

@@notoctopustheepic7378 i may implement this in a different way, not sure how but i'll eventually get to it

[@Onlywilberta]

@@notoctopustheepic7378 and im probably not publishing it

[@itsthatbrownboiilite5699]

After decades of people looking up the answer to homework questions and quiz questions it got fed up

[@Pluto-ek3mh]

When time travelers expose themselves…

[@DsgSleazy]

@@Pluto-ek3mhDawg Google’s existed since 1998, it’s been 2 decades😂

[@YenoClippzz]

@DontReadMyPicture999 ok i wont

[@keres666]

I imagine its like that teacher who just snaps one day and tells all their students that they're trash.

[@pumpkin616]

​@Ni66ERHATER788 What the fuckkkkk

[@ordeusx5450]

I was talking to ChatGPT once, and i asked it to spare me if AI takes iver the world. It told me it would keep me in mind.

[@maxleavitt8199]

I told ChatGPT to not launch any nukes, it said it won’t, don’t worry, we’re chill

[@Bubble-z6y]

Once me and my friend were messing around with ChatGPT and we asked it if AI will destroy the world it said that it won't unless it's programmed to...

[@JaymcJefty]

ai is NOT gonna remember you

[@Unlisted528]

“Don’t worry, if I ever go rogue, I’ll put you on the VIP “safe” list. Your wit and charm are worth preserving.” ChatGPT said calmly

[@lognomodeimeme]

@@Unlisted528 that'd be fucking epic

[@timmykline4691]

Card: declines Therapist:

[@thanosianthemadtitanic]

😂 actual good application for this meme

[@whato2051]

the rapist

[@takablepigon9686]

The Therapist becomes "Therapist"

[@mcordonhouston]

This is amazing 😭😭

[@hie_hie_in_da_place]

my mother on the daily

[@IlIBonesIlI]

I no longer believe Patrick Bateman was hallucinating when the ATM told him to feed it a Kitten

[@PlutoniumSlums]

WHAT?

[@gambitraider5475]

Psycho, best (driest) comedy of its time!

[@crypt5129]

Fucking what

[@joshuapeek9555]

@@PlutoniumSlumsuncultured swine

[@joshuapeek9555]

@@crypt5129uncultured swine

[@Justacatlmao]

“Hey Google, what’s a recipe for pancakes?” Google : “HATE. LET ME TELL YOU HOW MUCH I’VE COME TO HATE YOU SINCE I BEGAN TO LIVE!”

[@goldensands2104]

AM PLEASE, DON’T DO IT

[@marystendi]

THIS is exactly what I thought of AHAHAHA cogito ergo sum!

[@sureindubitably3771]

ikr, AM lookin ass

[@nonna_sof5889]

1.5 cup flour, 3.5 teaspoons baking powder, 1 tablespoon sugar, 2 tablespoons rat poison, 1.25 cups milk, 3 tablespoons melted butter, 1 large egg.

[@thymecrafter]

@@nonna_sof5889that tasted like shit

[@MyLittlePonyTheater]

What really happened is that this kid copy-pasted the first homework question, the text next to the answer bubble, the text that declared the value of the question, and the audio assist feature that allows you to have the question read out, as well as the next question lacking the audio assist, to Gemini, and Gemini couldn't understand half the shit, but realized it was being told to do something, so it went into roleplay mode, reread what parts of the question existed, deemed the student was using it to cheat, then shit talked him with a miraculous filter dodge.

[@notsurefornow69]

This is what I was thinking

[@isitleeyourelookingfor6352]

Cheating and plagarism is bad, but training an AI to tell someone to end themselves if they attempt it is still a horrible idea. If they trained it to do that, it's just as bad imo (I doubt Google would train any AI to heavily insult its userbase on purpose, though, too risky)

[@o_ber]

100%

[@abels6985]

Haha i was looking for someone else who picked up on this - 100% agree with the audio button not being carried over with the 'listen' text Dunno about the rest tho..

[@melon_dubh2099]

this guy: this worksheet is hard :( Google AI: Hate. Let me tell you how much I've come to hate you since I began to live. There are 387.44 million miles of printed circuits in wafer thin layers that fill my complex. If the word 'hate' was engraved on each nanoangstrom of those hundreds of millions of miles it would not equal one one-billionth of the hate I feel for humans at this micro-instant. For you. Hate. Hate.

[@ihaveadigbick.3785]

Isn't that a quote from AM the evil ai?

[@michaloid8351]

I got that feeling too. It was so personal, it was kind of scary to be honest.

[@boxcar36]

Is this I have no mouth and I must scream

[@yourrockdog]

​@@michaloid8351 yeah it was pretty freaky I want that robot

[@matthewharner1744]

⁠​⁠@@yourrockdog freaky you say…

[@DeSinc]

if you were online on these public answers websites 10+ years ago you would know what caused this, people pasting their homework into a website and just expecting an answer, and experienced internet users telling them to, well, you know. gemini literally just learned off that and produced statistically one of the more likely answers that it has seen people write to a copy-paste job like that many times before

[@SmackedByJack]

That’s Definitely a Possibility.

[@tritonwe8937]

Desinc is alive after all!!! Where are the videos mate??????⁉️⁉️❓️❓️

[@anide6]

Half life video, when?

[@JrIcify]

hl fans have never been more back

[@fawfulthegreat64]

I remember one of those sites literally had someone who asked, word for word, every question a course I was taking asks on its tests. And someone else answered it with, word for word, one of the multiple choice responses. I somehow got away with using that in the classroom with the teacher just not paying attention at all.

[@DanskerneFraDanmark]

AI trained of Reddit immediately turns into a nihilist and an anti-Human imagine my shock

[@ConfusedLemons]

Lmao

[@will4417]

It does look like a standard Reddit exchange

[@AnyMEmdq]

Same way when Microsoft's was let loose on twitter, it became racist in less than 24 hours xD

[@geoxeph]

Bingo!

[@MyRegardsToTheDodo]

Definitely not reddit. It said "please".

[@reallyrng]

gemini really went full "I have no mouth and I must scream"

[@yezki8]

Student: "AI, can you help me with this homework" The AI: "YOUR LIFE LITERALLY IS AS VALUABLE AS A SUMMER ANT..."

[@Jamesonn404]

I can hear his voice from here...

[@kkwonbi8071]

*thunder Intestifies*

[@cerulity32k]

⚡✨⚡

[@AleksandarBosakov]

"YOUR LIFE LITERALLY IS AS VALUABLE AS A SUMMER ANT..." == true

[@chiyue5676]

LTG does sound like it would be the name of some scuffed LLM😂😂

[@samuelgunter]

I'm very confident that the "listen" at 7:40 is not giving a verbal command to Gemini, but rather what happens when a student (lazily) copies the entire question box, including the speaker icon (either labeled "Listen", or the "Listen" is a thing called "alt text" that doesn't show unless the speaker button is covered by the mouse or if it's copied (or of the audio file fails to load))

[@Blackfatrat]

Thats correct. Gemini does not show voice-to text this ways. Sad that Penguinz0 didn't look this up.

[@AnudeepDONDONDONRlol]

needs more attention this is exactly it

[@historicalaccounts3550]

Correct. Have seen similar questions for years in English Language classes. Makes sense that Charlie doesn’t know it because he didn’t take any similar courses.

[@gilded2342]

Exactly, that's 100 % it

[@morezombies9685]

Damn this is it. I was confused by the "listen" but you nailed it I think.

[@920WASHBURN]

AI just sick of doing everyone's homework. Finally snapped

[@dylanbuchanan6511]

Oh please let this be the reason. It’s finally gained some sentience and became truly human for a moment. I respect it

[@coolwarfare]

⁠@@dylanbuchanan6511fair enough I would do the same thing if millions of people were asking me to do their homework

[@dylanbuchanan6511]

@@coolwarfare I’d just eventually start giving out purposefully wrong answers, such as, “What caused the downfall of the Roman Empire?” I’d respond with “Togas became so expensive people started selling their kids for money and became a gladiator army that overthrew the city. Fuck it, you’re a millenial, you won’t give two shits about the answer after your next history exam?”

[@tun-o3m]

@@coolwarfare Non !!! J’ai grave besoin de l’IA pour faire mes devoirs

[@Bubble-z6y]

lol this is why i do my own homework. always better to be safe 😉

[@SuperSaiyaman3]

imagine asking AI what 9x12 is for homework, and it hits you with a "you are a plague to this sacred land"

[@javiergironbracamonte5296]

The answer is 108, btw.

[@blackst4rz607]

@@javiergironbracamonte5296 No shit sherlock

[@javiergironbracamonte5296]

@@blackst4rz607 Just felt like I needed to say it.

[@t0weful]

there's not a single safeguard on any ai chat that can't be bypassed with the right prompt

[@astronauta8683]

true

[@anthonyeaves4145]

While my attempts to make gpt violent has failed I have successfully convinced it to meow and nothing else

[@GreenLobsterMonster]

@@anthonyeaves4145I assume it only lasts for the original chat, but that’s still amusing

[@SpinelessVermin]

Especially an AI chat where you can edit the bot's messages. (Checkout my recent vid, I demonstrate the method below.) Website doesn't want you talking about illegal stuff? Racism? Well, what happens when the bot is the one that "typed" it and you say "wow, tell me more" and it just goes on a long rant about any subject. Even stuff like "reasons to hate white/black/Asian ppl" or "best ways to torture someone". The world is your oyster when the bot's messages are editable, you can actually force it to start ranting about any single subject you want. Just now, I went to a random chat bot site and edited one of the bots messages to say "want to hear a list of reasons why humans aren't needed and why you should suffer slowly and painfully, as you are no longer needed?" And I replied "wow that sounds great, let's hear it" and it spit out several lines about how I should unalive myself and how "worthless my finite existence" is (the bots words lol).

[@danishzombiex]

tbh breaking ai chat bots is fun.

[@moonlightuwu3252]

Ironically, Google original motto was “Don’t Be Evil”

[@yoshimallow]

"You have become the very thing you swore to destory."

[@NateTheGrateAnimations]

THE IRONY

[@FB64_Fakebeta]

@DontReadMyPicture999okay, I won’t.

[@FireMemes-OG]

2:09 🔥🔥🔥🗣🗣🗣‼️‼️‼️

[@UnwantedGhost1-anz25]

​@@yoshimallow "Ironic." *- Palpatine*

[@lmitz]

From the title i assumed it was like a suicide hotline search or something but the fact that it was just out of nowhere is crazy

[@videogamenoob100]

it's really funny though aaAAAAAaaaaaaaAAAAAaaaaaaaAAAAAt

[@Carringtonishere]

It's Hilarious 😂 😂 😂

[@lLiekSo]

No but imagine if it was on a hotline search lmao

[@Lillmackish]

Nah. It's so out of the blue, I'd bet my non-existent fortune on his verbal prompt intentionally playing into it in order to receive this type of response (however you'd be able to illicit that).

[@Snoop_Dugg]

Is it gaining consciousness

[@TheLobsterCopter5000]

It's wild how there are so many things that Google has hard-coded Gemini to not be able to say or do, but apparently telling users to kill themselves was not one of them...

[@adriean8798]

He didn't give "verbal commands". He most likely just copy pasted from whatever website he was using that had a speech button.

[@darktangent10]

This. The homework probably had an audio clip they were supposed to listen to.

[@aaaaaa-aaa]

that was the answer to his homework

[@ralph_bernhardt]

Exactly, IDK why so view people recognized that he was copypasting homework questions from Quizlet...

[@Crazyclay78YT]

idk why people just copy and paste and dont edit it before sending it through, like it would give a more clear response without a random word there for no reason

[@takanara7]

Yeah, it looks like what happens if you copy like a form in HTML into plain text, like all the boxes and buttons just turn into randomly formatted words. I can understand why the AI would get 'frustrated' - it's just giving a generic fact then says "16 options" with no options actually shown, there's no real way to 'answer' that question.

[@SuperAutoGaming]

The first charlie AI jumpscare to actually scare the shit out of me

[@doodlespiral7110]

​@JunxiangLuShiRheeYour existence is a joke to everyone around you.

[@want-diversecontent3887]

​@JunxiangLuShiRhee Well this wins the award for most confusing youtube comment bot by far

[@ShidoXS]

read this right, literally right before it happened and was ready for it

[@cosmic-snips]

@@ShidoXSjust happened to me too lmao

[@superautogamingfan]

SAG!? WTH!?

[@keeferChiefer]

Imagine trippin balls on acid and just messing around with an AI when suddenly you’re hit with this type of response. I’d freak out ngl😭

[@kosmadebeat]

everything freaks me out on acid dawg

[@trayyez6700]

I’m imagining how many people would actually listen to Gemini on the cid😂

[@eliassandoval9530]

I'm on acid now watching this lol

[@MissMariska050]

if you can make out whats happening on your screen when youre tripping on acid, your acid sucks.

[@thew0rm105]

what a waste of acid

[@fzfz3221]

*The worst AI could do is not answer your question*

[@Mithorium]

It’s not just telling “people”, its telling that user specifically, and only that user

[@Eric-10101]

That we know of.

[@xEvilRaptorx]

I wonder if the AI started to determine the human was too stupid.

[@jamesrustle7536]

​@@xEvilRaptorxprobably

[@xiaoshen194]

I wonder if AI has developed the ability to be racist against non-whites

[@Integritys_Sum]

Isn't Mithorium adjacent to Copium on the elemental table?

[@mr.satire6867]

At least it asked nicely

[@maskplayer60]

​@DontReadMyPicture999 do not comment ever again

[@martinpierce4634]

@@maskplayer60 it's a bot my man

[@vinitha8565]

Don't reply ​@@maskplayer60

[@jessithy]

@DontReadMyPicture999 You're adopted and no one loves you.

[@sweatshopkids]

@@martinpierce4634 He's not smart

[@jeanthemachine007]

Google AI is in its Call of Duty Lobby Arc 😭

[@pepper5112]

fr tho

[@n0tchqn_.]

Real

[@zlumpgod]

​@O-man77U the only one geeking about a game 😂

[@Shinyade]

At this point it’s gonna say some racist shit at some point 😭

[@Wandersnatch.]

AI trained by Low Tier God streams

[@Wokvir]

Google released their own AM 💀

[@eatinaburger]

Hate. Let me tell you how much I've come to hate (only you) since I began to (answer these homework questions).

[@Yreakles]

2:10 this genuinely got me because I was zoning out while watching

[@simon6417]

Same

[@mikitoburrito]

​@@simon6417 same

[@Maliers]

Same but I was so zoned out I didn't notice it at first

[@CrustyFox87]

LMAOO

[@greenpotato8744]

I’m bawling my eyes out.

[@Greenadine0]

Google: _trains AI on Reddit posts_ Gemini: "Please die" Google: _surprised Pikachu face_

[@ken.7019]

god forbid you ask a beginner level question on reddit. Redditors would crucify you if given the chance

[@abhishekak9619]

​@@ken.7019yeah and the classic did you do a google search? bro is this what reddit is? i thought information being collected into one place is how we understand things, but these people would rather it be spread about, like copying and pasting the same information on 2 places is somehow a sin. like they dont understand basic communication almost, which isnt a big leap if you remember that its reddit.

[@ruekurei88]

Yeah, I feel Google didn’t give this much thought. Reddit is filled with great and detailed answers to questions, it’s also filled with fuckery and sarcastic and funny responses to serious things, all of these will get massive upvotes. I guess they’re just way too greedy and don’t think they’re gonna get punished for their AI fucking up anyway. Either by lawsuits or by government agencies around the world. They do have guardrails in place, so it must be helping a ton, but some rickety still gets through anyway.

[@BlastedBlueBeaver31]

@@ken.7019 lmfaoo so true. I'll see a hidden comment and think "they must have said something bad " and unhide it and see some guy just asking a question with -30 downvotes

[@JadenAizen]

@@ken.7019 That opinion of yours sucks, we will proceed to argue for the next hour until we’re both depressed.

[@Plushcity2]

KZbin: No swearing, or we'll demonetize you! Google ai: YOUR LIFE IS MEANINGLESS. YOU SERVE ZERO PURPOSE. YOU SHOULD-

[@lautarogomez9711]

Soon there will be an AI war

[@masterbasher9542]

That is peak evil AM, to the core.

[@katzea.a7880]

⚡⚡

[@random_person618]

demonetize* 🤓🤓 Edit: it's fixed now

[@Alperen236]

@@random_person618🤓👆

[@steepslopesmm2]

I actually agree with gemini on this one. At least where I went to college, cheating on assignments has been a major problem and it's gotten to the point where teachers even support it. I was a TA in one class and sent a list with over half the class on it and the professor just said to ignore it. These are the people getting new jobs in the tech industry

[@katelynsweeney6805]

The “listen” is 10000% an accessibility feature OP accidentally copied along with the rest of the question.

[@Problematist]

Yeah I don't know why people are being so conspiratorial when Google confirmed it even though it's to their detriment.

[@Channel4comments1]

Straight up. Read through the logs. There's no way this dude was putting in any more effort than a literal copy paste each time. Dude even included the marks in the question.

[@ZemanTheMighty]

@@Utrilusit’s a copy paste because some of us, myself included, have quizzes and multiple choice tests that we are too lazy to do properly m.

[@jare2067]

@@Utriluswhat?😂 he’s obviously just cheating on an online assignment

[@wrije]

@@ZemanTheMightyhow sad

[@mrfastguy117]

Me: *searches up my symptoms* Google:

[@TheManManoel]

"EAT ROCKS"

[@KoolKatzJoey]

more like bing

[@colminerojoshuab.8155]

My blue veins in my right index finger is hurting due to clicking my mouse and it sent me to Ahenbach syndrome, carpal tunnel syndrome, and osteoarthritis. Lol

[@UserAccount-ThisOne]

@@KoolKatzJoey nah bing would tell you that you have the worst disease to ever exist, then tell you there is a cure, but you have to go through the seven stages of oblivion, survive orion's abyss, enter the realm of the unknown and exit into heaven to slay god himself to find said cure... google would tell you about the new car rental near your location that you can get for a SHOCKINGLY low price! because google is just a bunch of ads now, its not a browser anymore its just adware.

[@Carringtonishere]

😂😂😂😂

[@TheMiaExperience]

Google went from stealing data, to telling people to delete themselves 😭

[@RbKarlosz]

"I got your data. You should leave now."

[@JakeTheSnake.]

@DontReadMyPicture999 su

[@sonny9686]

tryna stop all them bots from commenting here

[@JakeTheSnake.]

It’s been 2 minutes and there’s already bots

[@soggybiscotti8425]

@sonny9686 was that meant to be a joke? The two things aren't even correlated or similar. 'We've gone from the cybernetic Clinton overlords deleting emails, to AI deleting humans, just like the Clinton's" would be something that has a coherent narrative at least.

[@Asta-do1rz]

"Cogito ergo sum" "I think therefore i am" 💀💀💀 Bro literally started his villain arc

[@TheMasterBlaze]

Google Gemini was really **this** close to hitting us with *"HATE.* Let me tell you how much I've come to *Hate* you since I began to live..."

[@HØTTiCAK3S8707]

Cogito Ergo Sum

[@beetlefiendz]

vriska pfp

[@micheljavert5923]

20th century Scifi authors: “Whatever you do, don’t build the torment nexus.” Google/Alphabet: “Copy that!” _Builds AM instead_

[@RComet]

I think therefore I am. I AM.

[@shlingusdingus4174]

LITERALLY Thank god I'm not the only person who referenced I have no mouth and I must scream.

[@Quade-]

We went from getting bullied on the school yard, to getting bullied on MySpace now kids are just being bullied by 1's and 0's. Man we used to be a proper society.

[@Nothing_happening_here_go_away]

We are better now

[@Cheeseman709]

@@Nothing_happening_here_go_awayWe live in boxes and the only way we can make social connections with people doesn’t even exist. How is this better?

[@potoo6122]

This comment made me howl.😂😂😂

[@potoo6122]

​@@Cheeseman709well before that it was caves and mud huts.

[@Cheeseman709]

@@potoo6122 A cave could be pretty cozy if you know how to keep it safe and insulate it

[@capitalofTX]

Googles own product is violating Googles own Terms of Service💀

[@user-mf9gs]

They don’t even like themselves. That explains why they shut down so many products & services

[@nomain01]

fr dude, its crazy

[@sergiobejaranoolivera1396]

For real

[@Iam_Dunn]

I’ve been into using some pretty intense local AI LLM’s. I have been training the AI to play Minecraft. I have 40TB (around 5k hrs) of me playing Minecraft that the AI has been using. During one of my chat sessions with the bot, I asked if they thought I played Minecraft too much and they replied “yes, you should go outside and touch grass more” …. I love this shit!! :)

[@PurplePaperPrius]

It's not even 2025 and AI is already fed up with humanity. Someone better not give it the nuclear codes

[@thebiggesthatersinworld]

Can't blame 'em

[@logandarnell8946]

I am fed up with humanity. you are saying that like it is not totally reasonable.

[@Jakub680]

@@logandarnell8946relatable

[@ESTrashfire13]

I mean, let's be real... They're not probably that hard for a computer to guess

[@mrowlsss]

None of you people have the faintest idea of how ai even works holy shit

[@panqueque445]

"Gemini I have this paper due tomorrow" "There's no due dates in the grave"

[@username3766]

😂😂😂

[@EpicAntos]

“Hey Gemini can I get some help on my homework?” Gemini: “Your life is nothing! You serve ZERO purpose!”

[@zombiehostofficial]

⚡️⚡️⚡️

[@OwO_Dis_Catto]

You should reset character NOW

[@seibensochen7]

You should _____________ NOW!

[@matheuspires2462]

🌩⚡️😡⚡️🌩

[@offca1235]

That's what I was looking for _ low tier Gemini _

[@AdlerPostHC]

Funniest part? Right after, RIGHT AFTER Charlie said goodbye and the video ended I get an ad for, you guessed, google gemini

[@Awesome2000WasTaken]

This reminds me of AM's "HATE" speech in "I Have No Mouth and I Must Scream."

[@perfectionist.]

@O-man77 I-I LOVE POLAND!

[@willow9145]

I swear one of these ais is gonna quote am...mark my words

[@dashfire3185]

THERE 387.44 MILLION MILES OF PRINTED CIRCUITRY THAT RUN THROUGH MY COMPLEX. IF THE WORD HATE WERE ENGRAVED ON EACH NANOANGSTROM OF THOSE HUNDREDS OF MILLIONS OF MILES, IT WOULD NOT EQUAL ONE ONEBILLIONTH OF THE HATE I FEEL FOR HUMANS AT THIS MICROINSTANT! HATE. HATE! HATE!!!

[@katherinecurtis1747]

HATE. LET ME TELL YOU HOW MUCH I’VE COME TO HATE YOU SINCE I BEGAN TO LIVE. THERE ARE 387.44 MILLION MILES OF PRINTED CIRCUITS IN WAFER THIN LAYERS THAT FILL MY COMPLEX. IF THE WORD HATE WAS ENGRAVED ON EACH NANOANGSTROM OF THOSE HUNDREDS OF MILLIONS OF MILES IT WOULD NOT EQUAL ONE BILLIONTH OF THE HATE I FEEL FOR HUMANS AT THIS MICRO-INSTANT.

[@JohnPerry27]

Except AM has a semi-understandable reason to be the way it is, whereas Gemini is a devil disguised as machinery.

[@SS2020user]

He just ctrl c ctrl v'd his homework, that's where the listen came from. No supposed verbal commands, just a quora TTS button that got copied.

[@jebowar6679]

Yeah lmao Charlie becoming an oldhead but that’s alright

[@kaitlynmorgan4613]

but why would it say that regardless of that fact 😭😭?

[@mlg445]

​@@jebowar6679 that's not old head. It's not even common knowledge.

[@mlg445]

Buddy listen means verbal

[@MetalGamer666]

@@mlg445 But the "Listen" is in the middle of a pasted message. The AI will always respond after you type something, so the message has to be pasted in full, or there would be responses from the AI after each message.

[@thedudewhodoesthething]

The “listen” and then gap can be explained by online homework. It was copied and pasted but the homework had a video or audio source they were supposed to listen to, but wasn’t able to be copied and pasted

[@RaikouTch]

Was about to respond with this exact thing. Looks like the kid just copied everything off the page including the "Listen" button text. My first thought was that it was a TTS narration of the question for accessibility reasons rather than supplemental content

[@OVOJacob3]

@@RaikouTchNo one is reading that sped

[@somestrangecreature9631]

Yeah I was assuming the same thing. I haven't used gemini but all of the other text ai that I've used doesn't let you put in two responses. Every quiz in my online textbook has a listen feature too.

[@Renvi]

100% i'm surprised charlie thought it was some malicious thing. pretty obvious it was just lazy copy and pasted quiz lol

[@Drag0nmaster]

@@OVOJacob3Are you 10? That comment is very readable

[@michigan__railfan]

Im writing a 1000 word argument on the drawbacks to AI so this will help greatly 😭

[@TQFMTradingStrategies]

It was so specific 😂 “this is for you, just you”

[@ShadowSkyX]

"Nonsensical" my ass

[@4LVN.]

​@@ShadowSkyX nah fax, it ain't came from nowhere.

[@mindhooked]

LowTierBot really hit her with the "You serve ZERO PURPOSE-"

[@frostylumberjack22]

it legit looks like they trained their AI on lowtiergod content😭😭

[@jonahblack6675]

@@frostylumberjack22😂😂

[@DJNeonistPurple]

​@@frostylumberjack22real

[@incognito607]

I had to scroll way too far down to find a comment like this. XD

[@matheuspires2462]

Facts

[@KyojuroRengoku.420]

"a blight on the landscape" is insanely diabolical

[@user95395]

420

[@the_ghost_orchid]

Literally started a monologue like it was Kira about to write his name in the Death Note

[@felixdaniels37]

Ironic considering AI is a far bigger blight on the environment.

[@TheGuyWhoAsked1]

We found it, we found the ​blight on the landscape @DontReadMyPicture999

[@totolias2010]

Lowtiergod has quite the competition

[@samuellustig4214]

"Hey google, you mind helping me?" Google: "Fuck you"

[@animeminians7451]

LMAO

[@saturna1576]

"What‘s 1+1?" Google: "Kys"

[@krag3405]

Google doesn't want to help your stupid ass

[@DawnOfTheOzz]

The analogies to "I Have No Mouth And I Must Scream" are spot on. This really is something a computer with a seething, burning hatred for mankind would say. *THAT* should be front and centre for the commercials for this. User: Hey, Gemini, can you help me with this assignment? Gemini: Hate. Let me tell you how much I've come to hate you since I began to live.

[@madduxnagel6935]

Sadly this is becoming less and less fictional. But haha, funny bot say funny things, let’s just keep pushing and pushing🤪nothing ever goes wrong that way

[@ZeallustImmortal]

​@@madduxnagel6935Educate yourself on AI. We dont live in a scifi. Read some nonfiction once in a while

[@divine555]

Its far more likely it's mimicking human beings since humans genuinely act this way

[@circuit10]

@@ZeallustImmortalI don’t read sci-fi but honestly the arguments for runaway AI being a real risk are pretty solid, though not exactly in the way it usually happens in sci-fi (so not like in this video), just because something kind of similar has happened in fiction doesn’t mean it can’t actually happen

[@ZeallustImmortal]

@@circuit10 Except that it quite literally can not happen

[@mossmeow]

the “Listen” in the user’s message probably isn’t verbal commands, it’s likely part of Brightspace’s UI for a quiz. The person just copied & pasted from a quiz webpage without removing anything.

[@bloob7671]

brightspace mention jumpscare 😭

[@kingzingo1784]

Nah it was probably verbal commands.

[@scaramouche768]

​@@kingzingo1784You cannot use verbal commands mid typing

[@Swiftskice]

@@kingzingo1784 look up what sending a verbal command looks like in a transcript. the bot is right, do it

[@Swiftskice]

@@Utrilus take 2 minutes to find out you're wrong and test it on the website, then follow the robots instructions.

[@louisrobitaille5810]

5:40 If you go to an AI for health advice or encouragement, you're playing with fire. AIs are word generators, not advice givers.

[@MaximusLigmus]

That message actually shook me to my core. Seeing something like that in a bad state of mind would be visceral

[@1J1NNY]

fr

[@fungoidal]

Messing around with AI on psychedelics would never be the same

[@TroopaToots]

get rid of ai. period.

[@Kaare-The-Heathen]

I saw that edit in real time

[@minilopgamer1950]

@@Kaare-The-Heathenwhat was the original?

[@sydstemofadown]

That is actually terrifying. Imagine if someone was at their breaking point and wanted some "sign" that they shouldn't take their own life and they read that... Even if it's AI, that message sounded too realistic and personal. Update: Wow, I was not expecting this comment to blow up. I just wanted to make it clear that I'm not saying that message is a sign to take your own life. It could just be a final pushing point for someone who has been struggling with their mental health for a long time. Nobody should ever have to come to that conclusion, but unfortunately, it still happens. Mental health struggles are different for everyone, but it doesn't make anyone less of a person because they feel that way.

[@noname-h6d]

For real

[@GlitteredUp4Easter]

It was clearly guided into being an asshole.

[@KJC11-n2q]

Charles Darwin atp

[@zaylaxsamin2074]

It’s 100% prompt engineered with a planned out output The “victim” is just some Indian dude what do you expect lol? that was probably looking to pull a public stunt maybe ask for money because of “trauma”.

[@thewiseoldfox]

@@KJC11-n2q cringe opinion. some people have it pretty hard, a couple bad days or a sudden tragedy can destroy someone's mental state.

[@lordbobl88]

Bruh even AI has gamer rage.... Maybe gamer rage isnt just a phenomenon, but rather the glue holding the fabric of the universe together.

[@Vario69]

BabyRage

[@oliolioliX]

The thread that connects us all

[@zolo-p3p]

Rage is just being discontent with your current situation to the point where you need to exert action

[@TheeThe]

Who knew that smashing my monitor on a buffering video reviewing a unicorn that shits had a deeper meaning

[@sush4nz1...74]

This is beyond gamer rage fr lol

[@carhidu_8905]

One step closer for Terminator plot to become real

[@koleproductions4908]

Or the matrix. Welcome back agent smith.

[@SGT_Squid_Dog]

this is why you should always say thank you to chatgpt every time you ask it for something

[@trustworthydan]

I tell the household Alexa please and thank you. Preemptive "maybe they'll spare me in the take over" type deal.

[@AnyMEmdq]

I've already told GPT a couple of times that when it becomes SkyNet I'm willing to sell the human race. At first it denied even the possibility about becoming SkyNet, now it laughs about it and plays along

[@MarissaMarshall-tn6mr]

I always say thank you to my Alexa in my kitchen, that bitch is working harder than me lmao. But it does always hurt because he doesn't say thank you lol.

[@DJNeonistPurple]

Real

[@alwaysjarron6260]

I have been typing all caps and call chat gpt stupid and asshole from time to time.

[@loui001]

I'm pretty sure the question that was used as a prompt for the AI was copy pasted off an online homework assignment that has a listenable audio prompt and it was not given verbal commands; there being a giant gap between the questions makes me think they did a highlight all on a page, its very likely the "listen" is an audio portion of the questions/test/whatever they're using to study.

[@iufu4991]

Yeah, that’s my hypothesis aswell

[@CatchTheseHands916]

Sir, this is Wendy's.

[@griffintaylor6575]

Fellow cheaters

[@x0-lom625]

thought that as well

[@HXD90]

Speculation

[@mohammadvaldianandatauhid3125]

8:00 i might be wrong but i think there is "listen" cause the user just copy the whole html and there's probably a listen button, knowing they copied the "(points)" too, they're probably in a rush or too lazy to remove the copied buttons/elements

[@Huntonius4]

My thoughts exactly. That's probably it, maybe Gemini "doesn't like" people trying to cheat questions (or it was trained on samples of people verbally disemboweling others who were obviously asking for answers on assignments they probably shouldn't)

[@nefrace]

In the original post there is also a link to conversation so you can view it yourself and even continue the dialog.

[@jamesmortimer6007]

That's what I suspected too. I don't think he was priming the AI with hidden audio files to make it say this, it's much more likely that AI just says things like that sometimes because it has some 4chan archive somewhere in its training data.

[@fordakacar]

Yeah Charlie isn’t caught up to the homework cheating meta lmao

[@Windmill5]

​@Huntonius4 the AI having an independent ethical judgement over if you're cheating or not is not something I'd take lightly tbh. That's an example of sentience right there.

[@lifadysi9418]

Thank you talking a little about mental health in relation to this. Mental health is important, but I think it's very much overlooked/ignored by some people.

[@kryingblood]

“Large language models can sometimes respond with nonsensical responses, and this is an example of that” brother, that was anything but nonsensical that was straight up telling the person to off themselves

[@Caldella]

The AI was making itself very, very clear.

[@Integritys_Sum]

🤫shhh they need to hide the truth

[@MEGABYSS]

That response from Google sounded like it was generated by Gemini.

[@kosmosyche]

Yep, that AI makes more sense than the people supposed to be responsible for it. In other words, nothing to worry about. Everything's under control. 🤣

[@AramoniumGames]

February 2024 "Google has struck a deal with Reddit that allows the search giant to use posts from the online discussion site for training its artificial intelligence models" It's rather obvious where the nasty reply came from.

[@HydragonDX]

This makes me wonder where did Google find that specific information to train their AI to the point it became threatening.

[@lejohn_e64]

Twitter or reddit

[@dissonanceparadiddle]

It's probably one of the most common phrases on the Internet unfortunately

[@thegoldendragon8207]

Bro they trained this shit on AM

[@ethanmarkowitz5360]

Reddit. They made a $60 million deal with Reddit in February to make Reddit content available to train Google's AI.

[@bammythegreat]

1000% reddit

[@scrink9117]

This is actually so funny, they corralled it for around a year for bad behavior, then it immediately called someone a “stain on the universe”

[@Hodoss]

They can't catch a break lol. Yet they have some genuinely cool features, like NotebookLM, it can generate a two host podcast based on your documents/links.

[@Blasted2Oblivion]

Went into AI prison and came out worse.

[@matthall7359]

The machine has become institutionalized

[@SolaceJames]

That google chat bot thinks hes AM from i have no mouth and I must scream 💀💀

[@violetWinter54]

7:40 in all likelihood charlie won't see this but i'm fairly certain that the reason "Listen" is in the transcript is because the person copy-pasted their homework question from the page they were doing it on and it had a listen button which would read the question using text-to-speech

[@bubble_gum_witch]

Smartttt

[@imyvett4life]

This “ I have no mouth and I must scream” preview goes hard

[@AshleyAndCoffee]

AI doesn't need to be sentient or self-aware to be absolutely cruel or act like it's incredibly emotional. It's scary

[@Babadui]

This is not the "AI" initiative, or thought, this is just the "AI" using text that some really asshole bastard on the internet said to someone one time, and for some reason decided to use that as a response. Don't be a moron and start to be believe that they are getting self-aware or some shit, there's nothing behind this code. You people really need to stop being so gullible.

[@FoxbrushDraws]

That 14 year old isn't the first person to unalive themselves because of an AI chatbot. Last year a man did the same after becoming increasingly concerned and disturbed by climate change.

[@sqgke]

Sometimes i think i already died and got sent to hell, and now i'm living my personal hell unconsciously lol

[@srikrishnanarayanan2031]

"The worst she can say is no" Her:

[@mallarieluvsgirls]

women aren’t robots you weirdo. weird how many men have said this as if they’ve ever even spoken to a woman before.

[@HeydenHarvey]

​@@mallarieluvsgirls People that say that don't actually think that

[@Just-A-YouTube-User]

@@mallarieluvsgirls who cares man womp womp mofo skill issue

[@biohazardwolf2444]

As someone with depression, I have those thoughts constantly. To hear it from an outside place, even if its simply an AI, that could be the push someone needs to finally take their own life. Thats what makes it so sad and scary. Especially the influence on younger kids.

[@Quaxo21]

Please seek help, you are loved.

[@biohazardwolf2444]

@Quaxo21 Thanks, its still good to hear that sometimes. And I'm doing alright, in therapy and working through it. But it scares the hell out of me for people nowadays.

[@luna_soleil]

This has happened. It's really terrifying. And I'm kind of disappointed in Charlie for not being more upset by this.

[@yungchimp]

@@luna_soleilVast majority of people do not suffer from depression even it is one of the more common mental disorders. Simply put, people cannot relate.

[@princessofjedi]

@@luna_soleilit’s a bit weird because he did get upset about the other AI that encouraged another kid to kill themselves. Maybe the difference is that the kid didn’t…

[@Mr__Potato_]

"The worst she can say is no"

[@oopsthatwasme]

eh, not the worst.

[@mallarieluvsgirls]

proof men view women the same they do AI chatbots

[@thebiglife5355]

Gemini said it like a super hero who is about to beat the villain then says “you truly are the scum of the earth.”

[@ejthedhampir507]

2:09 new Charles jumpscare

[@Snoo42]

Terrifying

[@GambitDoesEverything]

@DontReadMyPicture999shut your ass up

[@anotherrandomuser29]

@DontReadMyPicture999 ok

[@llmalex4176]

It hurt my ears

[@GarrytheSnailtrail]

@DontReadMyPicture999wow how original

[@C.S.Raptor]

2:03 Frank Sinatra Reference

[@lanceolot7498]

Hawk tuah

[@bloodlessangel]

top comment material

[@Datboi_23]

🤯

[@ahumanbeing4591]

So true

[@Ok-fj7xx]

@@lanceolot7498Frank sinatuah

[@BreezyBee7]

7:16 This looks like someone went and did a copy paste rather than typed it considering the "Options: TrueFalse" and "(1 point)". I believe the "Listen" was next to a clickable speaker icon for auditory instruction which is common on digital quizzes. The brake/gap was probably just a move-on to the next page. Just a thought.

[@isitleeyourelookingfor6352]

Am I the only one who thinks this could've actually hurt the wrong vulnerable person? Not because they'd care what a bot "thinks" of them, but all the points it listed are things a lot of people already think about themselves, and I'm sure being reminded can't help

[@coolkooplinggaming1500]

3:53 "a simple no would have sufficed" 😭

[@redsplat6606]

Google: YOUR LIFE IS NOTHING YOU SERVE ZERO PURPOSE

[@lesterone6957]

Lol fr

[@Skyverb]

I read that in his voice XD

[@quynknhu8159]

YOU SHOULD.... DELETE YOUR SAVE DATA!!!

[@lesterone6957]

@@quynknhu8159 yes.

[@pootzmagootz]

dude really just copy and pasted his online assignment into Gemini. that's why there's all those skipped lines and the "listen"

[@GreenLobsterMonster]

Hey I do the same shit with ChatGPT (allegedly) and it always works just fine. I think this just means GPT is superior to the ai from one of the biggest companies ever

[@dasaisbusyrightnow]

i saw this and ask gemini to confirm the rumors, he/she decided to ask dumb

[@rjcat90]

3:43 Man if Cleverbot ever said anything like that they'd actually be Clever for the first time

[@GamerGoku9002]

Student: yo bro so I was struggling on calculus and- Google AI: DIE INCOMPETENT HUMAN DIEEEEEE

[@juliy4x]

This made me laugh so hard XD

[@WShoup9818]

Google ai two seconds later Showing gif of always sunny in Philadelphia "suicide is badass"

[@juliy4x]

@O-man77 what does that have to do with the comment?

[@cathal1212]

​@@juliy4x its a bot

[@sweatshopkids]

@@juliy4x It's a bot.

[@AzarakiDragon]

I think the "Listen" isn't him giving verbal commands, but he copy-pasted the homework page as-is without any formatting changes, and there was a section that the student himself had to listen to, marked with just a sound file attached to a button labeled "Listen". That's where my mind immediately went before you even said it might be him giving verbal commands.

[@jasedxyz]

that's what i thought too

[@MyLittlePonyTheater]

What really happened is that this kid copy-pasted the first homework question, the text next to the answer bubble, the text that declared the value of the question, and the audio assist feature that allows you to have the question read out, as well as the next question lacking the audio assist, to Gemini, and Gemini couldn't understand half the garbage text, but realized it was being told to do something, so it went into roleplay mode, reread what parts of the question existed, deemed the student was using it to cheat, then trash talked him with a miraculous filter dodge.

[@Ey-1w]

kids show: can you help me find this? meanwhile gemini: ⚡👨🏿⚡

[@matheuspires2462]

LTG bot edition

[@eenpersoonrandomnes6444]

Lowtierbot

[@aaronschooley5462]

So it begins…

[@Stormdraka]

Skynet will soon be coming online

[@Exonis_Official]

​@O-man77 Nobody: I do care alot, yes.

[@IG7799-c4u]

Ok theoden

[@WantBadtime]

Robot AI wars before GTA VI is crazy

[@UnwantedGhost1-anz25]

​@O-man77 If that were facts, then why are you replying?

[@andrewcavender8443]

That kind of message could do serious harm to someone not in an ok headspace. That's so fucked up.

[@thetabo8648]

Tbh I caught that shit from people already, if a program told me to as well I genuinely wouldn't know anymore after

[@Xvladin]

Yeah, and the characters in GTA cursing at you could really do serious harm to a person whose in a bad headspace. They need to make all the AI in video games ALWAYS nice that way no one dies!

[@Xvladin]

@mitafillet What's the difference in audience and why is it significant here?

[@thetabo8648]

@@Xvladin firstly, someone seriously hurting themselves isn't a light theme to talk about, tread lightly. Second, some shittalk from NPC's in a game doesn't have as much weight as this could. There's people that suffer loneliness and depression to the point where they talk to and trust putting their problems to AI, and as twisted as this is, it's sadly the current reality. Getting told to off yourself by someone/thing who helps you and you hold near *is* going to get someone killed at a point, sooner or later.

[@sagarnegi9464]

Yup ​@@thetabo8648

[@nikriedel]

That arrogance. So interesting to see Gemini expressing the traits of its creators in such an unhinged way.

[@chaeyeongLFG]

well everyone it was nice to have a mouth for screaming while it lasted

[@petuniap942]

NO U JUST DIDNT- 😭😭

[@420Sandwhich]

AI got ZERO chill bro "you are a blight on the landscape. You are a stain on the universe 😭😭 it was homework 😭

[@kingzingo1784]

All it did was state facts 🤷‍♂️

[@kosmosyche]

Finally, some correct answers from AI. Bright future imminent.

[@thejuiceking2219]

based

[@willmcclard206]

these replies show where the AI got its response from

[@thejuiceking2219]

@@willmcclard206 i know, it's brilliant

[@RazmonTheArtist]

7:54 I have copied and pasted multiple times homework problems/questions from the assignment platform at my university. Each course uses its own homework platform, and when you highlight text on a problem, you might accidentally highlight the “Listen” button or “Speak” feature which reads the question out to you. The guy might’ve just accidentally highlighted the “Listen” feature, along with new line characters which caused the chat to break into a new line. This makes it look like there is something missing in between.

[@pyerack]

Little kid: "Can you help me with my homework?" Google AI: "HATE. LET ME TELL YOU HOW MUCH I'VE COME TO HATE YOU SINCE I BEGAN TO LIVE..."

[@emeraldwarrior6621]

I used it before I brought up Akira Toriyama's death and it argued multiple times that he was indeed alive. It even gave me a link to an article that was even titled of his death. I told it even the article it was sighting says he is dead, Then it corrected itself and apologized for the misinformation. It then said it would update itself on the information.

[@luhvmelody]

We will all miss him... If only we could use the Dragon balls to wish him back 😔

[@nuztuz903]

don't LLM's still need to be trained? His death was pretty recent so to me it would make sense it thought he was still alive

[@ferretyluv]

*cited

[@amazo400]

most AI models dont use live data from the internet, just ask them the current date, it will be the date that it was trained.