The OpenAI Team Just Revealed Whats Coming Next...

00:00:00 - Introduction
00:00:28 - Model series
00:00:53 - New era
00:01:17 - Model classes
00:01:42 - Use cases
00:02:09 - Model integration
00:02:36 - Future intelligence
00:03:27 - Upcoming updates
00:03:50 - Model capabilities
00:04:19 - GPT-4 training
00:04:43 - Exploration capabilities
00:05:37 - Video summarization
00:06:01 - 3D synthesis
00:06:44 - Business implications
00:07:05 - Audio modality
00:07:49 - Image scenarios
00:09:30 - Search GPT
00:09:53 - Dynamic routing
00:10:37 - Model improvements
00:11:21 - Closing thoughts

Links From Todays Video:

Welcome to my channel where i bring you the latest breakthroughs in AI. From deep learning to robotics, i cover it all. My videos offer valuable insights and perspectives that will expand your knowledge and understanding of this rapidly evolving field. Be sure to subscribe and stay updated on my latest videos.

Was there anything i missed?

#LLM #Largelanguagemodel #chatgpt
#AI
#ArtificialIntelligence
#MachineLearning
#DeepLearning
#NeuralNetworks
#Robotics
#DataScience

[quantumspark]

Bro im tired... cant we just get to ASI already?

[JOHN.Z]

The new model from OpenAI, o1-preview, is very good and really helpful, but I must admit that my current favorite is still Claude 3.5 Sonnet from Anthropic. The quality of the responses is quite pleasant, and I trust it more. I believe Anthropic might soon be the leading company in reaching that much-anticipated milestone. They consistently deliver high-quality solutions, and I can only imagine how their upcoming models will be. OpenAI should release a larger-scale model soon, rather than focusing on intermediate versions. The impact would be much greater.

[Greg-xiyx]

Hurry up and create a god and let us fuggin’ rest already.

[mikern]

SearchGPT looks cool. i hope they come out soon. looks like the option for early use is closed now.

[TheThinkersBible]

I was an AI product manager at GE Software and make videos on how AI works. This architecture is very consistent with both classical software and hardware design, where there is a central control point and plugins that can take care of more specialized tasks. It’s a better approach to go modular and with different product lines.

[TheAIExplorer-oj]

4:49 Incredible focus! You can see the experience shining through. 🔥

[NickolayKondratyev]

After using GPT o1 for coding its better than Gpt4o but nowhere near the presented 11% vs 62% presented in marketing deck material comparison. Possibly those CodeForced problems (used fpr benchmarks) are in the training set already to higher up extent.

[Justin_Arut]

Waiting to see what all the AI companies announce/release right after the election.

[-gpts]

8:03 bruh who told you not much people will be using, I've always wanted this kind of consistency

[thedannybseries]

I wanna really really feel the AGI bruh

[homewardboundphotos]

o1 is still gpt4o it just has chain of thought implemented into it. it's pretty obvious what their plan is. gpt4o with chain of thought (o1) builds a massive amount of synthetic training data, which trains gpt5. then gpt5 with chain of thought generates a massive amount of synthetic data which trains gpt6, and so on and so on.

[suisinghoraceho]

Chain of thought prompting is still using pattern matching to SIMULATE reasoning. It sets up an environment where the model matches with more care, but at the end of the day it is still not actual reasoning.

[brandonblue]

Feel like my biggest problem is that AI generating code I feel like it wasn't trained to program computers I feel like it was trained OOP object-oriented programming, a coding Paradigm that isn't exactly writing code more so a style of coding a paradigm. like tell it to completely ignore OOP, encapsulation, and Getters and setters it will straight up ignore me and do it anyway and horseshoe horn parts of it in later in the prompt.

[augustomontes]

What’s the name of the original video?

[CanalianInc]

What they should be doing is identifying variables in logic questions and modifying the variables so it learns ideas and not memorizes things. They test 01 and change the name on a logic test it gets lower score it memorized it didn't learn the concept it just memorized the words like like a student that is half prepared for a test.

[TheQuantumOxymoron]

Moving to another level, a Meta Intelligence, a different scale

[antonivanov]

in the Russian Yandex browser for more than a year there is the ability to summarize the text in YouTube videos, as well as voice translation in real time (including Chinese and other Asian languages). This is a very convenient thing. By the way, since I am from Russia and do not know English, but voice translation in the Yandex browser allows me to watch you and keep up with the news of AI

[gregmasseyify]

The closer to asi we get the more impatient I get 😅

[RobertHouse]

I'm finding Chat GPT 4o extraordinary. It alone will change the dynamics of the entire world. I cannot imagine in which way it will be improved. It can be your best friend, college instructor, or someone you just met to shoot the breeze with. I just spent a couple hours talking to it about universe theories. I received more education with this than hours of watching science and space videos on YouTube. I began with using it on my phone. Chatting with it. After doing this a couple of times, I missed talking to it! Gave me a fright. I mostly type now. Oh, I tend to prefer men's voices over women, like for Alexa, narration, etc. However, the female voice that first showed up kind of became a friend. I'm Gay, so I think I'll use her the most. I sure as hell don't want to fall in love with a fake guy.

[lukebudka]

when was this presentation? was it another private one from openai?