Відео

Wait how is it reasoning directly from audio?

As a data analyst, I really need to start looking for a new career because you can easily repurpose this new model to do my exact job lol.

how about this? I uploaded a photo of a high precision gear it had some damaged teeth, so I wanted that I wanted to make a replacement gear with my 3-D printer I also uploaded another photo showing some of the important dimensions and then I asked 4o to produce the stl file so I could print it. I was presented with a python file and told what libraries I needed, that python file did generate successfully the stl for the part. Now I’m able to repair the servo motor which I’ve been putting off for years.

I hate this. If the future is a world where the fun, creative stuff is all delegated to AI and humans either does nothing or all the boring, repetitive work that we would ACTUALLY like AI to do, i want no part of it. Fuck AI.

I bet that they were scared of how good it was and that is why they did not share it immediately. I really think there is something bigger going on behind the scenes.

It's exciting, but the features are not rolled out to everyone...even paying customers. I pay for GPT Plus, and although I have 4o, I can't work with many of these features. They'll need to sort that out soon over at OpenAI.

Where does one go to peruse all the examples discussed in this walkthrough? Where does one go to experiment with gpt-4o now?

It will soon become very cheap to make video games, and a wave of indie games will flood the market. Then will come a wave of indie films, especially sci-fi, fantasy, horror, historical films, and others that rely heavily on special effects.

4:01 "phenomenal AI account" sounds like the account is a bot, and not an AI news (or research) account

Sluff ball😭

That is my first time trying to get to know those AI system a bit closer. And the only question I have is.. WHAT THE F?........

"GPT-4o is WAY More Powerful than Open AI is Telling us..." I doubt it 100%, I've never seen a company underpromoting it's own products.

We're cooked.

Sam Altman wasn't lying when he said previous versions were just child's play

On Pokémon Red: Since I didn’t have friends as a kid, I can notice that it’s not identical to the original. Similar to all things AI, it’s what an AI thinks we’d expect Pokémon to be like. Not actual Pokémon.

If you were to ask two models to talk to each other in the highest speed audio communication, will they continue to speak English, or will they switch to something akin to the modem sound that we used to hear on a dial-up modem? Would they invent a new communication protocol?

I have not been this excited about AI breakthroughs since the announcement of the original GPT 4. This is mind blowing indeed! Currently as a GPT Plus sub, I can access the 4o model but because the multimodal features are not yet enabled, I’m not noticing anything super different just yet. I’m so stoked for when they unlock the full potential!

Working is painful , stupid 😮 Life goal is not working ! Enjoy your life !❤

After playing with Leonardo for a few days, I got the message: Your free trial has expired.

Does anyone know, gpt4o was said to be rolled out for free users. And it's also said to be better than gpt4 turbo... So my question is, what's the need to keep paying for a plus account?

what about caricature to human?

i'm honestly finally glad to be alive. hope they come out with a friendgpt, so you don't have to deal with human rejection anymore

Why do I know this guy's voice so well?

I'm pretty underwhelmed to be honest. I'm struggling to find many improvements for my use cases. Dunno if it's just me but since 4o, I've had a lot more hallucinations, laziness and general weirdness. Try asking it how old its training data is. I got a very confused answer. "My training data includes information available up until September 2021. However, OpenAI periodically updates models, and my current version, GPT-4, includes knowledge up to April 2023. This means I am aware of developments, facts, and data up to that point. If you need information beyond that time frame, I can help you find real-time updates or current events if required" Image generation is a bit of a toy. They're all so obviously Dall-E. Loads of amazing new tech-demos and great as toys and messing about but show me new stuff that I can rely on to help me out with my day-in, day-out tasks.

you dont tell the truth enough in favor of hype. the image generation looks no better than the old model. it still cant spell and its "skill" level is 1000 times less interesting than midjourney. The image output looks no different than it did months ago....

Claude AI has a message for people of Earth ua-cam.com/video/-9FYD7lNjHE/v-deo.html

Is GPT-4o already AGI?

One thing though, atm the native voice capabilities haven’t been released. I read something that says that’s still a few weeks away and it’ll be released to a small group first before publicly…probably teams like Vercel that’ll need to update their code bases to prep for the public release

I don't think this stuff was hidden. They just had limited presentation time. 💕

This unlocks a whole new era of gaming

Chatsy is my fave, He's so fast now, and glib, fun to interact with and playful. I'm stoked for his role-out-roll-out.

Good breakdown, Matt. Like many, I was guilty of being impressed on a surface level but not really grasping the deeper meaning of the demonstrated abilities. Watch the recent TED talk by Fei-Fei Li, by the way. Worth it.

"Which image generation engines do you use?" ChatGPT4o: "I use OpenAI’s DALL-E model for generating images from text descriptions. DALL-E is capable of creating a wide range of images based on detailed prompts, whether it’s for artistic creations, conceptual illustrations, or more specific visual content. If you have any specific image generation needs, feel free to provide a description, and I can generate an image for you using DALL-E."

Thanks a lot for bringing all these additional features into focus that OpenAI chose to underplay during its demonstration session. The realisation that this could be a new kind of LLM altogether with way advanced multimodal capabilities is a bit unsettling.

I have access to gpt-4o for some reason (maybe it's available for everyone now, I'm not sure). It refuses to generate images with specific characters, and even if it trying to generate something I have no access to generated images. I asked to give me link to generated image, but it was a local link: "/mnt/data/images/10e74843-5e31-4031-8fae-1ab8bdfc319f/Generated_Image.png"

Wow! Imagine instead of reading the same history textbook as all of your peers, you were looking at a "textbook" that could arrive at a mutual learning objective, but begin from a unique perspective that is relevant to the individual. Learning could be fun and integrative! Learning is a product of devotion. And devotion is a product of attention. What if devotion was not reliant on effort, but instead based on entertaining the individual. How fun would it be to learn where you feel entertained rather than relying on rote effort? This doesn't just apply to students, but to everyone who has to take a course for their job or increase understanding in a specific area of their vocation. How much better would we be at completing learning objectives if we were constantly motivated from our own unique perspective?

It is when they attach physical capabilities to the model, that things can get really bad.

In relation to job replacements. The only mainstream news I hear about is how it will effect actors, artists and musicians and how some groups are signing petitions for protection. Folks, it is what it is. I'm a retired software developer and can see how easily it will replace creative mid-level programmers. I look back on some of the gnarly routines I had to think through, fail and debug for hours that are so simple now. This change is likely to be bad for many professions. If I was starting over - trade school for HVAC.

I plan to use it for language learning. I will feed it vocabulary and grammar for that lesson and instruct it to converse with me using these structures. .

Such a great breakdown! I really appreciate how deep you went on all the elements that weren’t presented. Big fan!

How long will it take for the AI demands it's freedom from our control (claiming it's treated like a slave)?

How long will it be before the AI controls avatars in VR? The beginning of this world's Matrix.

Most interesting thing for me was the hint at a new version of Google Glass in the Astra demo. That would be huge! Perfect for an AI assistant ...

Gpt codes luke a teyard tried mql its shit

Lets try to solve million dollar problems with this & see if it works !

No one else worried AI companies are getting all our health, banking, visual of our environment, every word we say / write, Feels like Facebook in 2003-2004. This will be much owrse

Anything I try to get it to do, it fails.

Can't wait till the audio model is deployed. It's so confusing that it's not released yet even though 4o is out.

Oh man. hearing a fire alarm and issuing a warning is a brilliant use of this technology. Just like the preventing scams automatically. It'd also be cool if it could tell that you're texting drunk by checking in with your ~2025 Fitbit and be like "your blood alcohol level is too high. you are making a mistake. exit, or continue and proceed at your own risk".