What's important about this new type of image generation that's happening with tokens rather than with diffusion, is that this is effectively reasoning in pixel space. Example: Ask it to draw a notepad with an empty tic-tac-toe, then tell it to make the first move, then you make a move, and so on. You can also do very impressive information-conserving translations, such as changing the drawing style, but also stuff l…
Is it able to break the usual failure modes of these models, that all clocks are at 10 min past two, or they can't produce images of people drawing with the left hand?
In my tests no, that's still not possible with the model unfortunately, but it feels like you have way more control with prompting over any previous model (stable diffusion/midjourney).
There is no hubris. The LLM technology has actually been existing for at least two decades, it's not some sudden breakthrough we suddenly discovered. And given the many billions of dollars it has sucked in, it's definitely a pile of crap. I have been a paying customer of Github Copilot for at least a year. Since the google search has been completely messed up, sometimes it can be useful to look up some cryptic error…
No, LLMs have not exited for two decades. What a stupid comment. Millions of people are spending thousands of dollars because they're recieving tens of thousands of dollars of value from it. Of course it's impossible to explain that to thickheaded dinosaurs on HN who think they're better than everyone and god's gift to IQ.
Please read more carefully.The LLM technology exists for at least two decades, well actually even more.You know the technology the LLMs are based on (neural networks, machine learning etc). I am not sure if after smartphones, the LLMs will now further impact intelligence of people like you. And take note: I have been one of the early adopters and I am actually paying for usage. My criticism comes from a realistic assesment of the actual value these tools provide, vs. what the marketing keeps promising (beyond trivial stuff like spinning up simple web apps). Oh by the way, I peeked into your comment history. I see you're one of those non-technical vibe-coder types. Well good luck with that mate and let us know how is your codebase doing in about a year (as someone else already warned you). And if you have any customers, make sure you arrange for a huge insurance coverage, you may need it.
I’ll just be happy with not everything having that over saturated cg/cartoon style that you cant prompt your way out of.
Is that an artifact of the training data? Where are all these original images with that cartoony look that it was trained on?
It's largely an artifact of classifier-free guidance used in diffusion models. It makes the image generation more closely follow the prompt but also makes everything look more saturated and extreme.
I wish AI companies would release new things once a year, like at CES or how Apple does it. This constant stream of releases and announcements feels like it's just for attention.
Apple held three big keynotes in 2024 plus multiple product announcements via press releases: May 7, 2024 - The “Let Loose” event, focusing on new iPads, including the iPad Pro with the M4 chip and the iPad Air with the M2 chip, along with the Apple Pencil Pro. June 10, 2024 - The Worldwide Developers Conference (WWDC) keynote, where Apple introduced iOS 18, macOS Sequoia, and other software updates, including Apple…
I really hadn't noticed all of those! I'm mostly intersted in Macs, so I probably subconsciusly filter out the other announcements. I guess I haven't developed that level of 'ignorance' towards AI yet."
I wish AI companies would release new things once a year, like at CES or how Apple does it. This constant stream of releases and announcements feels like it's just for attention.
Visual internet content is completely over. Pack it up
So I spent a good few hours investigating the current state of the art a few weeks ago. I would like to generate a collection of images for the art in a video game. It is incredibly difficult to develop an art style, then get the model to generate a collection of different images in that unique art style. I couldn't work out how to do it. I also couldn't work out how to illustrate the same characters or objects in di…
I didn’t mean art. I meant visual internet content of all kinds. Influencers promoting products, models, the “guy talking to a camera” genre, photos of landscapes, interviews, well-designed ads, anything that comes up on your instagram explore page; anything that has taken over feeds due to the trust coming from a human being behind it will become indistinguishable from slop. It’s not quite there yet but it’s close and undeniably coming soon
I mean I literally "tried it again" this morning, as a paying Copilot customer of 12 months, to the result I already described. And I do not want to "try it" - based on fluffy promises we've been hearing, it should "just work". Are you old enough to remember that phrase? It was a motto introduced by an engineering legend whose devices you're likely using every day. The reason why "everyone", including myself with 20+…
None of this stuff even existed 3 years ago and you're asking like we're talking about self-driving cars. What hubris. My god.
> None of this stuff even existed 3 years ago
Copilot has existed since 2021. "What hubris. My god." listen to yourself...
Consider using a better AI IDE platform than Copilot ... cursor, windsurf, cline, all great options that do much better than what you're describing. The underlying LLM capabilities also have advanced quite a bit in the past year.
Well I do not really use it that much to actually care, and don't really depend on AI, thankfully. If they did not mess up the google search, we wouldnt even need that crap at all. But that's not the main point. Even if I switched to cursor or windsurf - aren't they all using one of the same LLMs? (ChatGPT,Claude, whatever..). The issue is that the underlying general approach will never be accurate enough. There is a…
It sounds like you've made up your mind! Best of luck.
Can you find me a single official source from OpenAI that claims that GPT 4o is generating images pixel-by-pixel inside of the context window? There are lots of clues that this isn't happening (including the obvious upscaling call after the image is generated - but also the fact that the loading animation replays if you refresh the page - and also the fact that 4o claims it can't see any image tokens in its context w…
Just read the release post, or any other official documentation. https://openai.com/index/hello-gpt-4o/ Plenty was written about this at the time.
I read the post, and I can't see anything in the post which says that the model is not multi-modal, nor can I see anything in the post that suggests that the images are being processed in-context.
No, LLMs have not exited for two decades. What a stupid comment. Millions of people are spending thousands of dollars because they're recieving tens of thousands of dollars of value from it. Of course it's impossible to explain that to thickheaded dinosaurs on HN who think they're better than everyone and god's gift to IQ.
Please read more carefully.The LLM technology exists for at least two decades, well actually even more.You know the technology the LLMs are based on (neural networks, machine learning etc). I am not sure if after smartphones, the LLMs will now further impact intelligence of people like you. And take note: I have been one of the early adopters and I am actually paying for usage. My criticism comes from a realistic ass…