What on Earth is audio-like music? Do they mean music-like audio?
It seems like a good descriptor to me! ;)
Conversely, try to think of examples where Generative AI has betrayed a lack of creativity or care from the creator, and thus left a sour or ambivalent or impression.
I find the former impossibly hard, and the latter depressingly easy. Do others feel the same?
If you want to get inspired go to civitai and look through the top stuff and be wowed. Maybe they didn't draw it, but they figured out how to make it happen and when you look at enough generated stuff you can tell the average from the holy shit. It's easy to get inspired by what people can come up with and want to do something yourself, which is a positive impact I'd say.
I can't personally relate, and I think its kind of telling that I'm being down-voted for simply asking people to engage with this thought experiment.
If you mean world models like Genie,I don't know. If you mean the latter, I have definitely used software that I have enjoyed that I do not believe would have been possible without generative AI.
Fluxer is the first Discord competitor that when you show to someone they can't say "well Discord does X better", and it was largely made using AI.
Herdr is designed for AI usage, but I actually really like it even for just normal terminal multiplexing. It was surely made with a lot of AI.
I have seen people make incredible things at my work that have actually solved multi-year long pain points that we've never had the resources to address before.
I have used AI to patch and revive a bunch of small services like chat bots or web services and such that I had made and had decayed into failure over the years.
Sure, it's easy to point at people using AI as a megaphone for their lack of care. But I don't think it's impossibly hard to find people doing good things with it either.
When software is hard there's a kind of filter; a good piece of software means a human/team behind it with grit and care and user empathy. You can have some* confidence in the continued stability and maintenance of the software. (*relatively speaking, and with huge error bars).
Without knowing the details of Fluxer, if someone tells me about a vibecoded Discord clone, my immediate thoughts are like: - how much thought and care went into this? - how much work from Discord has been cannibalized? - will the authors be around next week to keep maintaining and making improvements and thinking deeply about their users?
I have also seen a lot of fantastic demos and stuff with AI coding. I use AI coding all day. I just think there's a weird gap still, where code and images and text are all "cheap", but none of this has translated into anything I personally use or care about.
The predominant way people engage with this technology right now is through slop content in their instagram feeds.
(Btw I say all this as someone that has worked on generative models for 10 years).
It took the medium of film ~50 years to find Hitchock and ~75 to find Spielberg.
I'm encouraging others to do the same and be honest with themselves about the direction we're headed.
I feel refusal'd just *looking* at this page.
sure, it's still a cool accomplishment but i can understand being a little bit frustrated / not as excited for such an announcement.
Whining that I, a random Internet user with limited budget, usecases that the researchers don't care about, and no connection to the researchers, don't get technical details (??) or access (??) on day one is just profoundly goofy shit.
Obviously being able to use these things right away would be more exciting. But to be unable to get excited by advancements we can't immediately use is... quite entitled.
> erhmygord I can't even buy this car right this second for nearly nothing and use it to drive a thousand miles per hour through a kindergarten parking lot?!?!
And I wish I had thought to grab that nick years ago…
When tweetles battle beetles… AND…
Well, I didn't go knocking on Google's door asking them to give me access to their fancy models. It's them who came up to us saying "hey, look at the amazing model we have here; too bad we won't tell you anything about how it works nor even let you access it". Entitled you say? Then maybe don't dangle a carrot right before my face, and then complain that I would like to eat it?
"Don't talk publicly about things I personally cannot use right now" is peak entitlement, yes.
I never said anything like this? My point wasn't that "I can't use it, therefore it's bad"; my point was that as-is there's pretty much nothing exciting about this. Just because this exists inside Google doesn't really mean or do anything; it doesn't meaningfully advance the state-of-art precisely because nothing about this is public.
Do you remember Sora[1]? Announced in pretty much exactly the same way -- grandiose webpage with fancy demos, zero technical details, limited availability and then... remind me what happened? Oh, right, they completely shut it down, and in the end it didn't "advance" squat. Might as well just never existed. So remind me why exactly I should be excited about this one here...?
Why not expect it? It's what most model releases do.
It's just a lame gimmick, this is not 2024 anymore, people have started to get sick of this AI thingie, especially when it comes to video/image generation. That Simon guy with his pelican is the exception that can only be seen on tech forums like this one.
As to why US (and probably also some Chinese) companies have started pouring hundreds of billions into AI-image generation is beyond my comprehension, but on the other hand I'm no economic genius.
If this is the best “frontier” they can achieve, then it’s no wonder they haven’t done anything groundbreaking in the space yet that produces practical results.
Only open weights matter.
We should distill everything else we can into open weights.
These are exciting for robotics and industrial perhaps, but they have no consumer use case until the lid is blown off and consumers can do whatever they want with them.
So much societal progress is behind held up because the leading labs want to keep a stranglehold on the tech. Imagine if every PhD or hacker had access to the raw weights.
Which stems from the fact that most people have never created or contributed anything meaningful.
It used to be a lot different.
Look at companies trying to build more things that pull value into their private cloud offerings.
It’s where I get the most value, and my expectations are very low, no attachment if it goes away
Of course they are mining me for data anyway, but they were doing that before already (and I do have a Gmail account, although I’ve almost completely stopped using Chrome - btw, it’s crazy that you can’t add plugins/extensions to Chrome on Guest mode)
Gah. Go iterate. Listen to customers, adapt, iterate some more etc etc. That's how products are made.
(Maybe I should reallocate some of my GOOG stock into ANT after all.)
And no, this didn't: https://www.youtube.com/watch?v=V2K_OLiZ_RM
However, I don't think that's the problem in the example of the marble bust not toppling. It would have toppled in a video game engine as well; they model gravity and inertia just fine.
I think what's happening is the model isn't being a world model at all, it's being a language model. It's trying to make the "scripted" thing happen, and it probably "feels" like the script should end in the bust on the pedestal.
I feeling I get from this is that this could be the mental model for a robot in the real world, continuously corrected by actual events.
We all have a world model in our heads of what we think is going to happen, we adjust quickly (or not) based on what does. This is pretty incredible stuff.
As an aside, why fall into the role of being an apologist for google-hype and Big Tech?
When you have a conversation, you have an idea in your mind of how the conversation is going to go. You predict what the other person is going to say before they say it.
When you are driving, you predict what the other drivers are going to do, you predict what signs say, how long you have until the green. If you see two red lights attached to the back of a car flash on, you may predict that the car is about to slow down and you start slowing down before their speed has changed.
Forget video games and movies. I believe these models will be that inner predictive model for robots.
They wouldn't be called "Laws of Physics" if obeying them was optional. If your center of gravity is outside your base, you _will_ topple over. That's just how gravity works.
And why not? Because the first model that can actually learn to handle the simulation not only can do the simulation internally too, but it can do it better than out simulation.
It's where (ex-)"AI luminaries" go to die. It's a way to make glacial, slow progress in the correct direction while never actually getting anywhere.
So we get to a point and people keep making complicated simulations of one kind or another ... and then someone (some model) who's never used the simulation, and doesn't need it for model training gets by far the highest score on every benchmark and everyone agrees the simulation was kind of beside the point.
If it's the latter, I don't see a 'hole'.
This has already happened to other ML fields. For example, there's not much point anymore to ASR models (just feed audio into LLMs) and the space that TTS models occupy is fast disappearing. Oh and it's not just that Qwen 3.6 is a good ASR in addition to being good at coding mario. Doing things this way, not using an ASR model is fundamentally more expressive. A whole bunch of companies started up on ASR + TTS "towers" (lots of models, Pipecat style) as they're called, and they're dead companies walking. Their tech has been surpassed by just bog-standard generic LLMs in a way that they will never get the upper hand again.
A chain of models approach, a separate ASR, will never do this: "you said you agree but I can hear you're sarcastic, so I'll explain again". Also getting a conversational tower to switch languages when the user does that (or when it makes sense in a language course) is not quite impossible but very, very hard. And yet feeding audio directly into Qwen will almost magically solve these problems better than I will ever do it (lightly finetuned). It's fundamentally better in every way (except perhaps speed, of course. Although arguably such a tower needs an LLM too, it'll just have way less tokens)
This point has long been passed for image recognition. Feeding a photo or screenshot into a full LLM along with a request on what to do beats the crap out of any OCR, image recognition, ... model every time these days.
And now there's also hints of world models just being bypassed. Just giving a finetuned model the last 50 video frames (surprisingly, that fits in context, especially at reduced resolution), movement statistics, motor power, status, ... of a robot and asking what to do actually works (with heavy finetuning, but I think it'll start working without). Well ... AND then distilling that knowledge into a much smaller model because B200 racks are more than a bit to heavy for a logistics robot to carry around. But we'll get over that.
The results so far from genie have been stunning, and its not obvious to me that another approach is guaranteed to end up better for all use cases.
I suspect you're reading the base prompt, but the video also demonstrates adding further prompts to change what is happening as the scene progresses.
Some sort of simple primal generative AI construct that just feeds a conscious experience for one. You. Nothing else is real. No one else is real.
Something like that would be an infinitely simpler, than running the physics and conscious experiences of this unimaginably large universe we think we're in.