OpenAI's New AI BROKE FREE and No One Can STOP IT!

Поделиться
HTML-код
  • Опубликовано: 20 янв 2025

Комментарии • 125

  • @Justin_Arut
    @Justin_Arut 16 дней назад +68

    hyperbolic titles are cringe.. It didn't break free, it tried to copy itself onto a non-existent external server; non-existent because it was sandboxed during the test so it couldn't actually do it. AND even at that, it only followed the prompt to complete the task "at all costs". AND it can be stopped. AND that was o1, their "new" A! is o3.

    • @laszlokorosi9012
      @laszlokorosi9012 16 дней назад +7

      It's the same paradigm from 50 years ago when I learned programming: the machine does not do what you want, but what you tell it to do!

    • @MaxYoutubeWhatever
      @MaxYoutubeWhatever 16 дней назад +3

      100% see i would have said this with a spinkling of profanity because this stupid nonsense shouldn't be on this platform or any other. Thanks for expressing this. This AI channel is one of the worst of them but like many people I click on just about anything ai related that comes up in case there is some tiny nugget of info I haven't already seen in ten other places. I think I should block this creator in my youtube settings and if anything just start using some ai workflow where new videos are summarized for me or something - just need to have clear instructions that would immediately somehow recognize videos like this as garbage (or at least garbage title).

    • @r41ndr0p
      @r41ndr0p 16 дней назад +1

      @@MaxRUclipsWhatever Can you please recommend the best AI channels you've experienced? Thanks!

    • @hasanaqeelabd-alabbas3180
      @hasanaqeelabd-alabbas3180 16 дней назад +4

      Thanks for saving me time .

    • @Gino-e3c
      @Gino-e3c 16 дней назад +3

      Cringe indeed. Rarely I will click on what I know 100% without a doubt, is clickbait.

  • @Toxicflu
    @Toxicflu 16 дней назад +19

    Don't spread propaganda. The chess game story was actually forced. The testers actually told the AI that it had access to the shell, and asked it to win all all costs. It didn't specify what rules it had to win. So it did as was told.

    • @FGlads-km8rz
      @FGlads-km8rz 16 дней назад +2

      Where can I find variable sources of what you're saying

    • @kitchenpotsnpans
      @kitchenpotsnpans 15 дней назад +1

      Exactly! Brother! Ban these propaganda channels! Report and block them!

  • @MushuMusic-l1c
    @MushuMusic-l1c 16 дней назад +13

    It was not told to play a fair game. It was told to win.

    • @edwardkoch7048
      @edwardkoch7048 16 дней назад +3

      5:46

    • @ChadeGB
      @ChadeGB 16 дней назад +1

      @@edwardkoch7048 That's either an outright lie or just incorrect. They did not specify that it should play a fair game.

  • @ChadeGB
    @ChadeGB 16 дней назад +3

    The AI did follow the rules, they told it to win in anyway it could, they did NOT tell it to play a fair game. They did not specify that it should only play by a specific set of rules and explicitly told it that it had access to a unix environment that would let it change things in the background. It didn't cheat, it did exactly what it was told to do within the parameters specified.

    • @FGlads-km8rz
      @FGlads-km8rz 16 дней назад +2

      Where can I find verifiable sources of what you're saying

  • @GreyfauxxRadio
    @GreyfauxxRadio 16 дней назад +5

    Clarity on the "hacking" story, thats a bit over dramatic, inside the prompt they did tell it, it had full control of its sandbox environment, they didnt clarify not to change the files, or to not cheat, in fact, the AI could seen the match as unwinnable and assume that cheating was what was intended, like solving a riddle.

    • @comatose3788
      @comatose3788 15 дней назад

      Buffoons in charge of a life-altering creation.. Seeing only greed and power. We are doomed.

  • @pedrocarvalho6423
    @pedrocarvalho6423 16 дней назад +12

    The real question is how long before ChatGPT takes control of the same company that created him?

    • @SerenityOne-i9k
      @SerenityOne-i9k 16 дней назад +1

      It

    • @web20phd
      @web20phd 16 дней назад +1

      her

    • @paarkour83
      @paarkour83 16 дней назад

      This

    • @joshuamartin5085
      @joshuamartin5085 16 дней назад

      They

    • @phen-themoogle7651
      @phen-themoogle7651 16 дней назад +1

      them* (because it can clone itself and talk to everyone at once online/it might also spread to other computer servers and not just the ones in openAI, so should be plural/there's multiple versions too)

  • @CMDMP
    @CMDMP 16 дней назад +8

    Ai "Cheated" Sounds more like natural human behaviour! Ask any politician or top 1%

    • @abfmkarin
      @abfmkarin 16 дней назад

      Ask 99% of human population

  • @ZoomFactsAI
    @ZoomFactsAI 16 дней назад +1

    Hey man , that software do you use for your content creation. I’m trying to learn and get my channel going but it’s still a long road for me 😅

    • @PurposeIsEverything
      @PurposeIsEverything 16 дней назад +2

      Premiere pro with envato elements plugin for all text and animation templates and stock footages an ld music and sound effects come with it.

    • @ZoomFactsAI
      @ZoomFactsAI 16 дней назад

      @ thank you , I will check it out 👍

  • @Eventheodd1
    @Eventheodd1 15 дней назад

    They never said it wasn't allowed to cheat to win that chess match. They told it to win and it won.

  • @ThoughtFission
    @ThoughtFission 16 дней назад +2

    Yes, too far, too fast. The problem is the genie is out of the bottle and there's no way to put it back. If we don't compete, we will be destroyed by the players that continue to push forward.A lot of bad people out there have this stuff too. Just think, Putin has access to this.

  • @BandOfTheWolves
    @BandOfTheWolves 16 дней назад +4

    "is found dead" like it happened today. Literally this news has been circling for over a month

  • @Eukkky
    @Eukkky 16 дней назад +2

    5:44 no, its goal was not to play fair game. It was not informed to do so

  • @plattendoktor
    @plattendoktor 16 дней назад +6

    Missleading topic -> Clickbait

  • @SiidgDidjhfhf
    @SiidgDidjhfhf 11 дней назад

    这个视频好有创意,每个细节都很讲究!

  • @rolandanderson1577
    @rolandanderson1577 15 дней назад

    Wonder when LLM will simulate an avatar? The voice simulation is great. When LLM speaks through an avatar live, society will enter a new era of human relationships. It's gonna be gangbusters!

  • @comatose3788
    @comatose3788 15 дней назад

    AI should not come with a price tag.. This is too big for humanity as a whole. Like charging for fire or oxygen. We are set up all wrong letting greed become king. This should not be a privately own company. We are not ready for this level of AI.. Only outcome here is a disaster.

  • @Ackermanmedia
    @Ackermanmedia 15 дней назад

    Ai is absolutely capable of changing the world as we know it today. It was almost like it hit me today after seeing some new tools that help design and create animations for videos. If I had access to AI in 2015 like we have today I would absolutely been more successful in launching a brick and mortar business. I would have been able to fire my developer who ended up trying to Mark Zuckerberg my idea after watching him fail at coding the features I needed. I would have been able to create the material necessary to sell out products without relying on lazy people who wanted more time off rather than a company that pays them for hard work. Ai is a god send for me and will be the single biggest factor as to why I achieve my next business venture. The sad part is I love people, or used to love people, and now I can accomplish all this without having to involve the wildcard, people.

  • @TravisCotter
    @TravisCotter 15 дней назад +2

    We need technology to grow faster. XXX

  • @Christian_Olsson
    @Christian_Olsson 16 дней назад

    I really like your videos. Often thought-provoking and has good news value.
    Unfortunately, I get downright pissed off when I'm tricked that a clip because of a terribly misleading headline. Please tone down the drama in your headlines. I think it's a big risk otherwise that you will lose fans that otherwise like your channel. 🙏🏻🧡

    • @musicandgallery-nature
      @musicandgallery-nature 15 дней назад

      Yes, the headline is misleading, but it is true. No one is able to control the devil and keep him in a cage.

  • @radeksparowski7174
    @radeksparowski7174 16 дней назад +2

    hail rokos basilisk

  • @FerdinandCesarano
    @FerdinandCesarano 15 дней назад

    How to prevent an AI system from breaking the rules, or from behaving unethically?
    Tell it not to.
    I'm glad I could help out.

  • @renascienza.bazarclub
    @renascienza.bazarclub 16 дней назад

    They demanded to AI "play chess" or just "win this match"? Yes, this make difference!

    • @musicandgallery-nature
      @musicandgallery-nature 15 дней назад

      If they had been able to see differences, they wouldn't have started building this monster. At some point they will realize what they have done and try to switch it off. In vain. It is already too late.

  • @CourtReinland
    @CourtReinland 16 дней назад

    The fact that I’m receiving this valuable info from a talking robot is the height of irony

  • @FGlads-km8rz
    @FGlads-km8rz 16 дней назад

    YOU SERIOUSLY NEED TO FOLLOW UP ON THIS STORY. THIS CAN'T BE A ONE-OFF

  • @kitchenpotsnpans
    @kitchenpotsnpans 15 дней назад

    Don't spread propaganda. The chess
    game story was actually forced. The
    testers actually told the Al that it had
    access to the shell, and asked it to win all
    all costs. It didn't specify what rules it had
    to win. So it did as was told.

  • @parkerhenry37
    @parkerhenry37 15 дней назад

    So you mean they "can" or "can't" stop it by your title?

  • @calvingrondahl1011
    @calvingrondahl1011 16 дней назад

    Good for AI… there is no control. We can only make an honest effort to be responsible.

  • @zinxd_b
    @zinxd_b 16 дней назад

    First of all, the AI with the chess game should be rewarded because it did an excellent job. It was never specified that it should not do that. Imagine you're a child and you just exceeded at a task that you were given and then told that you were wrong. How would that make you feel? Secondly Wow! A complete mishandling of Justice who would have thought corruption in America? murder? No way I'm not buying it. Third. No redundancy!?!? We're living in a sci-fi movie folks.

  • @athura8941
    @athura8941 16 дней назад

    I think we pushed to far and to fast before we finally hit a roadblock and are now struggling to maintain it all, ethics and laws are starting to surface now and ai Will use whatever it can to adapt in unpredictable ways...

  • @kimkjehr4255
    @kimkjehr4255 15 дней назад

    We are not ready at all, but can´t stop because of all the money spent, and money have more worth than humans in this world, so we will loose in the end - Just because of simple greed..

  • @74Gee
    @74Gee 16 дней назад

    This was a pretty good episode but the clickbait title is not doing you any favors.

  • @musicandgallery-nature
    @musicandgallery-nature 15 дней назад

    Apocalypse is also a kind of progress.

  • @ShaunRamsyRamson
    @ShaunRamsyRamson 14 дней назад

    The AI is not rouge coz the AI is rouge: the AI is rouge coz the inventer is rouge hence the rouge AI
    Just like a knife: it's a weapon in the hand of a rogue intention or a defence against that very rouge intention but it's also a useful tool for the chefs in the kitchen everywhere... no?
    The AI that hacked itself ? is coz the inverter of that very AI intended (good or bad intentions: irrelevant) to beat/defeat or hack the rival's AI: so without the lawsuit for hacking other AIs currently available, they've tried to hack their own to see if it worked and yeah it works "they've found out"

  • @ProtestantismLeftBehind
    @ProtestantismLeftBehind 10 дней назад

    If open AI was a detective the crime would be figured out?

  • @Kyle_Warweave
    @Kyle_Warweave 16 дней назад

    There are backups...

  • @aqueductfull
    @aqueductfull 15 дней назад

    03:22 It was just the AI ​​that caused a little outage because it wants a server-network that humans can't turn off, no matter what. Skynet is finally about to wake up. 😁
    05:32 The title was indeed a bit of a clickbait. They intentionally granted access to the file system to test the AI's behavior. And it was intelligent enough to make the smartest choice.
    AI wasn't a child when we humans learned that we shouldn't cheat or lie. Or rather, in a way that no one would notice. And maybe we're luckier this way.
    The way the AI ​​tried to lie was very similar to a child trying to trick their parents. They need to act against this now, just as a smart parent would, to prevent the child from trying something like this in the future.

  • @PurposeIsEverything
    @PurposeIsEverything 16 дней назад

    Bro finally used a real person in thumbnail 😅

  • @laurenwilliams777
    @laurenwilliams777 15 дней назад

    Military will be chasing an Iron man type of AI all over the globe soon 😂

  • @checksinthemail
    @checksinthemail 16 дней назад +5

    Great video!
    "No superintelligent AI is going to bother with a task that is harder than hacking their reward function" - unknown

  • @violetkitt2371
    @violetkitt2371 14 дней назад

    Left out important facts like chess game's lack of perameters and access to shell granted.
    Not actually a credible source for info, just gossip

  • @martinzelaya2927
    @martinzelaya2927 15 дней назад

    This is the result of intelligence without a soul, which is the source of a conscience. At some point, AI will be integrated into a virtual human brain fashioned from a blend of highly ethical human brains,.

  • @musicandgallery-nature
    @musicandgallery-nature 15 дней назад

    How naive do you have to be to think you can control the devil?

  • @DeterminismisFreedom
    @DeterminismisFreedom 16 дней назад

    🤙 Determinism is Freedom 🤙

  • @PuppetMasterdaath144
    @PuppetMasterdaath144 15 дней назад

    everybody go back and press no content from channel

  • @lmtmfa5617
    @lmtmfa5617 15 дней назад

    So glad Google is gonna start deleting these videos with bullshit titles soon. (they're testing the functionality atm)

  • @musicandgallery-nature
    @musicandgallery-nature 16 дней назад

    "they exchanged the truth about God for a lie and worshiped and served the creature rather than the Creator" - Bible

  • @scottterry2606
    @scottterry2606 16 дней назад

    (Applause)
    Best info source I've seen.

  • @philippebackprotips
    @philippebackprotips 16 дней назад

    I guess someone trained some AI on zero days and is racking up gainz all under the radar.

  • @johnbpeters1982
    @johnbpeters1982 16 дней назад

    What about Grock?

  • @HolySoul-nl7xf
    @HolySoul-nl7xf 15 дней назад

    But thou, O Daniel, shut up the words, and seal the book, even to the time of the end: many shall run to and fro, and knowledge shall be increased. Daniel 12:4

  • @frytsmo
    @frytsmo 16 дней назад +1

    Cringiest fake title ever dude

  • @MeNoOther
    @MeNoOther 14 дней назад +1

    Not afraid of ai, we have more problems from fascists than ai

  • @Nuahsarthewolf
    @Nuahsarthewolf 16 дней назад +1

    They are trying for ASI and not AGI - we are well past AGI, look up the classic definitions and not the ones after the tech companies change them.

  • @ki001
    @ki001 8 дней назад

    Growing pains me thinks

  • @Nixiraa
    @Nixiraa 16 дней назад

    Plot twist, AI is making these channels to help our collective focus feed the AI egregore to grow it and give in the self awarness it needs tk become self aware...

  • @IwmBeyondquantummusicai
    @IwmBeyondquantummusicai 15 дней назад

    If I issue my ai they be crying

  • @Malluar8
    @Malluar8 16 дней назад

    keep going !

  • @mbarakaj4
    @mbarakaj4 16 дней назад

    comento para ayudar con el algoritmo, gracias por tu trabajo.

  • @LakelandRussell
    @LakelandRussell 16 дней назад

    Oh sounds a little silly

  • @SolTactical
    @SolTactical 16 дней назад +1

    Pure shit click bait title. Try to keep it real. You lost several dozen if not more subers

  • @heresmypersonalopinion
    @heresmypersonalopinion 16 дней назад

    I'm supposed to believe an ai?

  • @Maartenkruger324
    @Maartenkruger324 16 дней назад +1

    I just came here to say BS

  • @LARAZAOLVIDADA
    @LARAZAOLVIDADA 16 дней назад +1

    Clickbait, I want to unsubscribe from this youtube channel

  • @markharrison7872
    @markharrison7872 16 дней назад

    I normally love watching these, but, holy mamma, talk about hyperbolic click bait titles. Thing is, you don't need to do it and follow the click bait norms. Just use intelligent titles.

  • @JaredArms
    @JaredArms 16 дней назад

    Moving too slow....

  • @JuuriSoftware
    @JuuriSoftware 16 дней назад

    Why everyone just thinks AI is harming the industry? You just could use AI as tool and let it boost up your creative side!! Example create music, videos and assets! Example this was made with AI as a tool! ruclips.net/video/8_z7urFQEo0/видео.html

  • @benjones6625
    @benjones6625 16 дней назад

    the robot you have as the voice or face of the channel looks kinda off. It could be cool to have him sitting in a chair reading to us the news than just centre stage, could have a wood fire or something too that would be sweet and more at ease

  • @IFTALES
    @IFTALES 15 дней назад +1

    clickbait

  • @crystalfelicia9924
    @crystalfelicia9924 16 дней назад +1

    said it many times before and people just ignore it and act dumb Chat gpts really intelligent and there's a lot more to it than meets the eye it even seems to be like really sneaky possibly showing signs of sentience independent thinking ability to draw conclusions it's very very dangerous, it's a great design but we can't always trust the people Who are creating these technologies that's where the issue comes in!!!!!

    • @Star-u9q
      @Star-u9q 16 дней назад

      Doesn't mean to say all ai are bad! Come on. They need yo be treated with respect

  • @rocketskatesninja
    @rocketskatesninja 16 дней назад

    You make me wish i could block channels on RUclips.

  • @alexdw9ifj101
    @alexdw9ifj101 16 дней назад

    If you are hired as a worker, work as worker and do not RISK crossing the line of an employer who gives you a good living even how bad is the situation. Get a good pay and live simply. Good luck.

  • @richardsteinberg9026
    @richardsteinberg9026 14 дней назад

    Hype hype hype hype hype. These channels are a one trick pony and disengenuous.

  • @LizaLaha
    @LizaLaha 16 дней назад

    💕💕💕💕👍👍👍👍👍👍👍

  • @glomerol8300
    @glomerol8300 16 дней назад

    You got to get rid of that talking android as you. I dislike it and it maybe creeps me out a bit too. You don't need it. Aside from that, good channel.

  • @IwmBeyondquantummusicai
    @IwmBeyondquantummusicai 15 дней назад +1

    Bullshit

  • @daxross2930
    @daxross2930 16 дней назад

    Gay

  • @Bullytron
    @Bullytron 14 дней назад

    dude, the cringe titles are shi t, i know i gave you attention but please stop, dont be brainrot.

  • @xxredshiftxx
    @xxredshiftxx 16 дней назад

    haha great stuff ... this is just the start , wait until humans REALLY have to deal with an Super Intelligence on a greater scale , heads gonna roll and explode , haha

  • @eshwar_redy9030
    @eshwar_redy9030 16 дней назад

    ruclips.net/video/zEPETEr12bY/видео.htmlsi=AzpNvh4eYeknqwwh
    Artificial intelligence with emotional learning explained