5 MINUTES AGO: OpenAI Just Released GPT-o1 the Most Powerful AI Model Yet

Рет қаралды 159,130

Күн бұрын

Пікірлер: 72

@workingTchr Ай бұрын

How about simply TESTING it for us? Ask GPT-4 a tough question and then ask o1 the same question and compare the results. Don't take OpenAI's word for it. If you have a GPT-4 account you can ask o1 30 questions a month.

@prodromosregalides3402 Ай бұрын

It's better. I have already started doing my little testing. I am not sure it's about safety ,though. Gpt 01 does NOT fail simple logic stuff, whereas gpt 4o, sometimes(which is not very rarely) fails. How is this safe for 4 and 4o?

@HowToShopGreen Ай бұрын

Not sure you can do the test you ask and come to a conclusion that one is better over the other. It is like asking a physics question to one person that has no physics knowledge and then asking one person with a high level of physics knowledge the same physics question.

@workingTchr Ай бұрын

@@prodromosregalides3402 Just curious, why are you concerned about "safety"? Are you thinking of yourself of or others?

@workingTchr Ай бұрын

@@HowToShopGreen You'd have to pick a topic you already have knowledge of. But a person could research a niche area of a topic they are generally familiar with and that few people really know (or care!) about. Some arcane legal or technical question for example.

@Sem5626 Ай бұрын

AI boys are almost as bad as crypto bros with the false advertising the model is still dumb AF goes in circles suggesting bad things, it told me to remove security measures from my web app so I could achieve what I wanted to do it keeps on suggesting the same broken code over and over and over again, I got it to do it 28 times after telling it to keep in its memory that it was a bad solution but what are we supposed to expect, it's just predictive text based on answers from stack overflow which 90% of them are bad

@calvingrondahl1011 Ай бұрын

Patience is required… calm down people. Good for Open AI.

@richardphillipslivemusic Ай бұрын

2024: The AI Real Boom! 2026: The Year of the AGI.

@bassplayer807 Ай бұрын

I think it will be 2030 before we see the first example of AGI and I’m being optimistic, Ray Kurzweil said 2029 is what his prediction was, but even if he’s just 5 years off ( early) that’s scary.

@piteshbhanushali1140 Ай бұрын

2032 year agi..

@vincent_hall Ай бұрын

0:21 I love the implication that AI couldn't solve the problem of putting two puzzle pieces together.

@hatasmuveletinspiracio Ай бұрын

I think soon we will be able to generate videos like: "Make me a tutorial where Elon Musk talks about the science of quantum mechanics."

@dnasredine Ай бұрын

Wow, this is mind-blowing! 🤯 The 'o1' model sounds like it's about to redefine what we thought AI could do. PhD-level intelligence in physics and math? 🚀 Can't wait to see how this changes everything from research to everyday tech. OpenAI keeps pushing the boundaries, and I’m here for it! 🔥💡

@wildstorm74 Ай бұрын

Well, hopefully by end of next year. AGI will be here for everyday use…anyone will have a “tool” that have knowledge of the best of the best researchers out there, in anything ones interested in. Meaning one doesn't have to PhP in a field to create, share and improve on something.

@dnasredine Ай бұрын

@@wildstorm74 Totally agree! The idea of having AGI at our fingertips is both exciting and game-changing. Imagine the possibilities when anyone can tap into that level of knowledge and creativity. 🚀 The future is looking wild, and I can't wait to see how this tech empowers everyday innovators to make breakthroughs! 🌟

@desiraybelcher8371 Ай бұрын

Doctor know it alls AI Chanel will likely be the first to compare different AI

@OndřejMa9999 Ай бұрын

It’ll be interesting to see how it performs compared to previous models. I’m curious about the potential applications and how it might change things moving forward....

@Brian-ti7xm Ай бұрын

*I suspect it's a complex search engine. At times I felt like ai was genuinely thinking at others I saw a repetitive response and non immersive chatbot experience happen. It should be tested to see if chargpt is solving problems or if it's finding answers already available online*

@ClemMambulasa-m2q Ай бұрын

It solves

@Guyverman01 Ай бұрын

They need to make the next model better at being a creative assistant.

@wildstorm74 Ай бұрын

I do agree with you, GPT 4/copilot just needs to be better. The problem with that is, people in creative space apart-ally don't like their work to be used like that. Mainly because they don't get paid for it, as they so at a least. In my opinion, those tech companies lawers should turn around and “If those content creators/ news reports want to get paid for their work, put it behind a pay wall. That way, if those tech companies want their work. They have pay for just like everyone else does. In the agreement of pay, the buyer can use it how they wish too. After all, they paid for it.

@wildstorm74 Ай бұрын

…….also if they don't want there work to be use like that, even if someone pays for it. They shouldn't be working in that field to begin with.🙄😒

@beeilve Ай бұрын

@1:52 This entire video was AI generated. The creator doesn't even care enough to correct obvious mistakes.

@ClemMambulasa-m2q Ай бұрын

Haha

@RobertMurray-wk5ib Ай бұрын

The gods are gonna kill us. I can feel that weird tingling on back of neck.

@adamz1988 Ай бұрын

It's still a preview version If you ask it how many "r" there are in "strawberry" it will say 2

@Ryan-pj5cx Ай бұрын

Chat GPT - please tell me how to transform the data I have from the US Census Bureau to the proper model form to build models projecting down total category spending for each consumer in the USA and what model is best (including hyper parameters) with that data given the computers the business has available and production deadlines. .... ... .... yeah. that's what I thought.

@NickDrinksWater Ай бұрын

In 5 minutes: the next most powerful ai model yet!

@amanuelzewdie8462 Ай бұрын

When will we get agi?

@wildstorm74 Ай бұрын

Well, going by OpenAi. The “strawberry" model meant to be AGI baby. Meaning as it gets older, it will get better to the point it could match the best of the best in every field.

@wildstorm74 Ай бұрын

As for when, well they by end of next year….but wouldnt hole them to that date though. After all, we don't to be to upset about it….to point. Like how Elon and companies failed to keep to those deadlines.

@lukashardy4471 Ай бұрын

Can you afford over a thousand dollars a month?

@mydreams8092 Ай бұрын

14th Oct 2026

@Ratgodx Ай бұрын

@@lukashardy4471what’s the point of that question?

@cameronroman506 Ай бұрын

How many lies do humans have to stack up before the whole stack of lies falls over?

@ummekulsummasuma3575 Ай бұрын

That's mind blowing. From Bangladesh

@vaiklo Ай бұрын

Sounds like a boring software ad of the late seventies. And AI remains a simple big data cruncher, not an "intelligence".

@grumpyguss Ай бұрын

You're referring to "sentience" which isn't the same thing as "intelligence" which is defined as "the ability to acquire and apply knowledge and skills". When you ask AI tools complex questions about math or physics it has the ability to acquire and apply the skills that then solves the problem. It literally loads into itself incredibly advanced math abilities to do so. It seems to have textbook intelligence but it's true it doesn't have sentience. That being said, sentience isn't particularly useful for the vast majority of AI workloads.

@vaiklo Ай бұрын

@@grumpyguss I refer to the ability to solve complex problems or make decisions whose outcome benefits the actor without calculating risks or chances in money. Does that fit in your mind?? We are NO MORE in the past century, in which we are balancing digits to a abstract answer of what we would get as a payment from others, We are in a reality that puts the answer BEFORE our question now. But that are in no way complex problems to heel our world.

@vincent_hall Ай бұрын

Sweet, when I can use AI at the same ability as PhD students, I'll get a load of research done. Despite these claims, I'm guessing we're not at that stage yet. Either way, I'll still want PhD students. Among other students.

@tajabdullah.malaysia Ай бұрын

It cannot articulate innovative thinking, then to manipulate

@Nobestudy Ай бұрын

Nobestudy intelligence is breakthrough

@vincent_hall Ай бұрын

Wrong title. It's not "GPT-o1" it's "OpenAI o1".

@godned74 Ай бұрын

I hope so because up to this point gpt 4 cant code my complex projects anymore then I can.

@CMDRScotty Ай бұрын

The people I have watched to test it say it is more like a grad student than a PhD student

@kyliegufoster5544 Ай бұрын

Taylor Steven Rodriguez Nancy Jones Jose

@AinHab Ай бұрын

idk if it has Phd level intelligence , also this whole video is just AI generated ,would be better if you actually put thought out and give realistic review rather than reading out prompts that too from AI tools guys i also did a review of GPT-o1 , im just starting out would appreciate your feedback

@ShannonJosephGlomb Ай бұрын

Long live 1o ❤❤❤❤❤❤❤

@tristiada Ай бұрын

5 minutes ago but video length is 11:41😮

@Mahmood.Khokhar Ай бұрын

After GPT03and 04, again going backwards and making GPT 01?

@wshyangify Ай бұрын

It was GPT 3, 4, 4o, and now O1

@drishrat6334 Ай бұрын

o1 Vs o1 preview Vs o1 mini???

@The101Point1 Ай бұрын

It's terrible I'm still using GPT 4 or 4o

@cm-a-jivheshchoudhari9418 Ай бұрын

So much yapping so much theory without providing the actual necessary details, this video is definitely made using AI

@yaelbj Ай бұрын

😂

@sgatea74 Ай бұрын

Selling more strawberries 😃

@chad8061 Ай бұрын

5 minutes ago, but the video was made a day ago?..

@edwardk3 Ай бұрын

I hope it is developed for safety. Women could be harmed by the things it says

@vision6D Ай бұрын

Have done simple math tests on o1 and was happy with the results. Now I did the same tests on 4o and got better results any thoughts? The difference is on this probability example: : In a room of 30 people, what is the probability that at least two people share the same birthday? Assume 365 days in a year and no leap years. o1 result 44.9% 4 o result 71% After asking model o1 if the result was really 44.9%, it corrected it's reasoning and provided a new result of 70.6% If interested here is a link in which o1 does the initial reasoning and calculation: kzbin.info/www/bejne/aH2sfHdnmryXesU Have a great day.