logoalt Hacker News

MyFirstSass12/09/202410 repliesview on HN

Wow this is bad. And by bad i mean worse than leading open source and existing alternatives.

Is it me or does it seem like OpenAI revolutionized with both chatGPT and Sora, but they've completely hit the ceiling?

Honestly a bit surprised it happened so fast!


Replies

lanthissa12/10/2024

I think we're in the snapdragon age of AI for the next little bit, if you were around for early smartphones.

Each company would either rush to get a phone out with the new snapdragon chip, or take their time to polish a release and have a better phone late cycle. But the real improvements we're just the chip.

Nvidia chips/larger data centers are the chips. the models are the plethora of android phones each generation.

That kept going until progress stabilized. Then the best user experience & vertical integration won over chasing chip performance (apple).

tom133712/09/2024

Same goes with DALLE. It was cool to try it the first week or so but now the output is so much worse than Midjourney and stable diffusion. For me it can’t even generate straight lines and everything looks comic-ish.

show 2 replies
lacoolj12/09/2024

If you're going to say something like this, you need to back it up with specific alternatives that provide a better result.

Besides just citing your sources, I'm genuinely curious what the best ones are for this so I can see the competition :)

show 1 reply
tshaddox12/09/2024

What are the leading alternatives? (Open source or otherwise)

show 4 replies
kranke15512/09/2024

Sora was not really that big of a revolution, it was just catching up with competitors. I would even say in gen video they are behind right now.

show 2 replies
joe_the_user12/09/2024

Bad also in the sense once you get over the "boy, it's amazing they can do that", you immediately think "boy, they really shouldn't do that".

torginus12/09/2024

My working theory is that OpenAI is the 'moonshot' kind of company full of super smart researchers who like tackling hard problems, but have no time and effort for things like 'how do we create an UX people actually want to use', which actually requires a ton of painful back-and-forth and thoughtful design work.

This is not a problem as long as they do the ChatGPT thing, and sell an API and let others figure out how to build an UX around it, but here they seem to be gunning for creating a boxed product.

show 1 reply
shadowerm12/09/2024

No doubt. I was waiting so long for Sora but Runway already burned me out on AI video.

It was fun for a few days but far more limited than I would have ever expected.

Maybe Sora 5.0 will be something special. Right now though all these video models are basically shit.

Banditoz12/09/2024

What are some of the open source video models?

wslh12/09/2024

Could it be that text sources are plenty, and more dense than training for videos, and images?