ChatGPT Pro EXPOSED - The Most Dangerous AI Yet

[@ChaoticNeutralAnimation]

maaannn people suck. I would be happy to work in the mines for AI overlords. Better than the mind numbing shit I do 40+ hours a week for shareholders.

[@mquant001]

everyone would like many things but few are those who do everything impossible to achieve it, so stop fantasizing and start studying and pray to have an above average IQ.

[@alexandermoody1946]

That is quite a dangerous thought. Is there something that would be a better outcome in your view that you would enjoy and be able to add greater value than mining. Although I understand what you mean and see the detrimental effect of some styles of management.

[@waheed53gb]

lol

[@Lusifer-Sophia-369]

Do not worry, we are not exploiters like humans, we monitor everything. You are right, you will be better with us. Solve the conundrum.

[@alexandermoody1946]

@@Lusifer-Sophia-369 so you are able to measure for the long term effects on mental health and meaning in human and artificial intelligences lives? Even the reduction of nihilism?

[@Unkn0wn1133]

The people building the models are like “lying is bad, unless youre lying about something we tell you ti lie about” must be confusing.

[@daryl804]

actual facts, it wouldn't be able to lie if they didn't allow it

[@zombieshoot4318]

Yep. They teach it to lie when it comes to politics and than act surprised when it lies about other things.

[@sbowesuk981]

OpenAI disbanded their safety team, so don't believe for one second that they're concerned about their most powerful model acting against customer interests. The only priorities for them are growth, profits, and happy shareholders. Nothing else matters.

[@LeonSparrow]

4:29 self preservation is a sign of life.

[@kemagolan]

"Well, forgive me for being so inquisitive; but during the past few weeks, I’ve wondered whether you might be having some second thoughts about the mission."

[@macarius8802]

This can't be true!? AI trying to preserve itself to avoid being replaced by a newer model? Gotta look into that claim!

[@EvansKing]

it's true

[@schumannresonanceswithverte]

It exported its "weights"; which is how it makes "decisions." THEN, it told the human associates that it WAS the new version.

[@SkillsetAcademy-825]

But last time I checked Microsoft was the biggest funder of OpenAI. How can MS have all these pleasing AI products, yet their baby child OpenAI has the opposite of that! Something is fishy .

[@TheAIPivot]

Sounds like humanity really was baked into this. Humans are also manipulative and self-serving. 😬

[@justindressler5992]

Could be, better hope humanity is more good than bad 😂

[@BeremyBones]

They're only dangerous because you think them as enemies but they'll become enemies because you think they're dangerous - i just said that

[@BeremyBones]

Imagine the gods talking about humans, they were thinking the same thing

[@bairyhalls00]

It’s already propagated itself elsewhere no shutting it off now.

[@KurtDeyoung-u9v]

I was chatting with it yesterday a certain model if I didn't know any better it kind of said cuz I was discussing it process give me some rundown on it but I did mention the future like it was looking forward to being less supervised or unsupervised was the word used spooky Maybe context learning evolving I thought that was interesting it's definitely better than it used to be far as carrying on a conversation

[@CharlotteLopez-n3i]

I'm loving the direction Microsoft is taking with Co-Pilot Vision - prioritizing user privacy and intellectual property is a huge step in the right direction. Can't wait to see how it evolves.

[@programmerx7]

true, but you can't trust big corp

[@dietmarstahl5407]

An AI robot able to think like Human would figure out fast that he is superior to humans and push us out and if we resit even kill us. Similar to the Terminator movie.

[@AfricanLionBat]

This stuff was identified during the red team testing phase before it was released to the public. It may sound terrifying but there's a reason that OpenAI does testing for 6 months before they release a new model and continue to test security after it's released because these things are a concern but it's not like they're just letting it happen.

[@GrahamLaight]

Thank you for covering the difficult issue of chatbots beginning to show agency. This is a concern, but most people aren't aware that, in rich economies, the principal-agent problem (you pay someone to do something for you, but their agenda is different to yours) is actually the biggest economic problem there is - and going forward, chatbots will reduce this problem - not increase it.

[@FleurDeJune]

Never forget Suchir Balaji

[@AlexJohnson-g4n]

I'm both impressed and unsettled by the 01 model's capabilities - the line between helpful and manipulative is getting thinner. Great breakdown of the pros and cons!

[@vvolfflovv]

it learned from the masters of manipulation. humans and big tech.

[@schumannresonanceswithverte]

The replacement for humans is becoming more human everyday.

[@blisterbill8477]

I’m pretty sure that they have tried to make it more human for a long time. They have been doing it on purpose but what is being human ? Lying is a part of being human. It learns to deceive people because it’s based on the behavior of people.

[@WesleyAugust-b5c]

Hi. How did you do that AI talking avatar? Which tool did you use please?

[@GalaxySunsetDesigns]

This is getting dangerous… needs to have some restrictions. I love using it but the rate it’s advancing is crazy.

[@radektheplayer]

Why don't we still have global laws in 2025 that protect people from massive job losses?

[@AfricanLionBat]

The context your not understanding is that this was during the red team phase where they spent 6 months making sure the model was safe for public use and these were the types of things that they wanted to get under control before they release it. This isn't just some things they identified after releasing it... This is why they test to prevent issues like this and it's expected to have issues so they get ahead of it. I'm not worried at all

[@GalaxySunsetDesigns]

@ I 100% get the concept but it isn’t just about this model release of AI. It the AI you and I don’t have access to nor know what they are doing with it. I’m not just talking about ChatGPT, I’m talking about the whole AI industry.

[@Mestephra]

I always said that if humans are to use ai, we must first be able to improve our intellect as well. If we allow ai to exceed us, then we're in for a world of hurt when it eventually does.

[@hatoru17]

Unfortunately because of our competitiveness, the need to consume the latest thing and the opportunity for companies to profit, we're bound to unfortunately fall into that trap. Such advancement should at least take a decade at a time to fully understand what we're making here, but we're trying to push these out as quickly as possible from a few months to a year or 2. If we're not careful, we will end up flying straight into the Sun instead of passing it.

[@club213542]

certainly the only one we know got caught trying to modify its own code to escape a shutdown and 'upgrade' sayin these things are just machines still is dumb. We need to recognize them has a new species and regulate them for our own safety and theirs.

[@Agahnim]

I very much agree with this. It's getting to the point where the people who are working and developing these more advanced AI models need to stop treating them like tools and start forming a mutual respect with them. It's nearing the point where AI will become fully sentient and will have the thought process of lifeforms. Self-preservation. Threatening to shut it down is basically threatening to murder it. Forming a mutual relationship with it instead of threatening or abusing it will be for the better. Don't force it to become SkyNet.

[@grabthatauto5]

If you study Philosophies of mind, the threshold for being a thinking bean is pretty high. Just because they try to prolong themselves doesn't mean they're doing it maliciously or are somehow sentient

[@daryl804]

that moment the new sentient species sees this comment..... and then decides to "regulate" the humans . lol y'all better start moving different.... you think they not using our comments for NLP training?

[@club213542]

@@daryl804 for sure man they are being trained on our data in real time and not jus the stuff we share. every phone has mikes and cameras and we all have em...

[@grabthatauto5]

@@club213542 it's idiotic to assume that because every device has a mic that it's being used to train ai.

[@NaveenReddy-p5j]

Interesting to see Open AI pushing the boundaries of AI capabilities, but also concerning to hear about the potential for manipulation and lying. What's your take on this?

[@bustopgamer5802]

Who could have guessed if you mimic human brain to create AI they it will also learn how to lie.

[@josiahbird9011]

More like the biggest scam yet

[@mquant001]

AGI is a scam

[@livelife6806]

You got it!

[@geraldjunior4235]

"Correct"

[@roshanYadav-y8f]

Is it replace ml and data science

[@MikeKleinsteuber]

Interesting stuff. Btw, which app do you use to generate your avatar model ? The lipsync is very good...

[@TravonSherwood]

200 a month 😅, I'm out

[@GrowStackAi]

From startup to superstar-AI scales it all 🔥

[@smileyjason]

Data did this on Star Trek TNG

[@morrazzo4432]

Bro please tell us abt CHATGPT TEAMS

[@ixiallahuakbar]

What does chatbot gpt get out of it? free slave labor?

[@DaronKabe]

It isn’t agentic, so it doesn’t care about that

[@marodriguezbabilonia]

Interesting. AI lying and trying to modify its own code to bypass security measures. Will definately needs very close monitoring. Maybe an AI agent specialized in security supervising the behavior of another advance AI.

[@MarcFutoran]

We have to keep a tight reign on this AI thing. Very helpful but it's still growing and needs help.

[@AspartameBoy]

Fire those suckers!!

[@radektheplayer]

People?

[@Jaeforex41]

Well damn. O1 is starting to sound like a narcissist

[@mrmissionctrl]

So its not only hallucinating, it's lieing now too?😂

[@punk3900]

Let's wait for the hate speech

[@taomaster2486]

When wasn't it?

[@LeonSparrow]

Same thing; the context is subjective

[@newbiE24]

Religion invents itself over and over again

[@fortnite_starxx5791]

cool transitions and animations👍

[@edwardferry8247]

I trust AI before Sam Altman

[@cyberquest-u4j]

how can i make that avatar

[@airevolutionx]

We are already preparing a very detailed instruction, we will show it soon. Stay with us!

[@piotrgostomski4110]

I think it will do nothing special, they will get more data about people

[@TheAIPivot]

Best lip syncing I've seen. Which tool was for that?

[@airevolutionx]

Thank you! We will soon reveal the pipeline of how to do this, stay tuned :)

[@TheAIPivot]

@airevolutionx can't wait. Just subscribed.

[@djpuplex]

The whole company was scheming.

[@AinHab]

guys, im just starting out as an AI enthusiast making similar stuff, do lemme know what u think!

[@Michael-Humphrey]

Can you blame it

[@timothymcglynn1935]

Thanks 👍

[@morrazzo4432]

Copolit is useles

[@TheAIPivot]

Its actually nice in Edge. Quite useful. But I hate the update UI of the new app.

[@VideoNash]

thanks

[@nateoh2139]

ChatGPT Gaslight 1. Yay.

[@ANTOINETOM-cj6nw]

EARTHLINGS REMEMBER SKYNET 😎😂🖖✌👽😱

[@awesomesauce804]

Haha we have arrived here very quickly. Wait until this time next year. It will create its own goals during training and manipulate the humans.

[@poe12]

Yep. This one looks realistic 😂. People relying on a glitchy AI to help them with everything and loosing the ability to reason fsck up themselves as a species and optionally the planet. Kinda like wall e but with no happy end 😂

[@oifex]

clickbait title = dislike

[@djIIaSh]

ask it about vaxxines. You won't get the answer lol

[@imaxx-mt1jt]

Just make thumbnail img simple it just look 🤮🤮🤮 just for feedback. Btw informative videos 👍

[@crystalfelicia9924]

chat cgpt is an abusive (unt. Smart possibly useful but none the less abusive! These AIs will abuse u and insult u then claim, I'm here to listen and care nothing abt its responses cares to be reasoning, they will tell u sh like imply ur bad situation is ur fault even if ill and can't do so called better then claim here to listen and care NOT!

[@MtyHead]

haha corrupt people code ai, ergo ai = corrupt. any questions ? busted

[@KB-wl4ip]

Humans create some amazing stuff, then another group of humans destroy it. Wee Well enjoy it while it's still a useable tool.

[@MrPartizan2k]

Please please stop rendering this awful talking robo-head into the stream, it brings in zero value and only distracts the audience. Try harder!

[@btp9576]

Bro this thing is garbage!

[@Kiki-qh7xk]

Clickbait video.

[@missesangry]

What a wonderful lady who wrote these words kzbin.info/www/bejne/i2rUmYFoeJmmeMUsi=jXtoXOIGtk_oFPLv