Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Imagen Video: high definition video generation with diffusion models

imagen.research.google

Post reply on HN
Print view View original

441–450 of 500 posts

Re: Imagen Video: high definition video generation with diffusion models

#441

Post by ricardobeat »

PinkMilkshake wrote: post #252
alcover wrote: post #152

Earlier quoted context omitted.

Get better [...] so that you stand out Please bear with me but this kind of advice is often a bit puzzling to me. I suppose you don't know the person you're replying to, so I read your advice as a general one - useful to anyone in the parent's position. If you were close to her, it would make sense to help her 'stand out' in detriment - logically - to strangers in her field. But here you're kind of helping every read…

I [...] don't [...] like [...] helping a stranger. That's not very nice. The world would be a better place if we helped strangers more.

That's a good one, but if you read his comment thoroughly, it is about the illogicality of 'everyone standing out', not the 'get better' part.

Re: Imagen Video: high definition video generation with diffusion models

#443

Post by roca »

ALittleLight wrote: post #402
kajecounterhack wrote: post #374

Earlier quoted context omitted.

I'm sorry you feel so cynical about this. It's absolutely true that Google is profit-seeking, that these models are very expensive to build, and that if there's a competitive advantage to be had, Google should probably try to retain it. But even with that all being true, real people (typically some thoughtful researchers) build these models. And my point is: _there really are ethical reasons to keep large generative…

I'm curious what ethical reasons you think require that new technology only be used in secret and without oversight by trillion dollar companies. This is supposed to be AI safety? "Do whatever you want, just make sure you conceal the results and impede progress and understanding." What reasons necessitate keeping image or video generation models private that wouldn't also argue for keeping animation software or pictu…

Keeping GPT-3 behind an API lets OpenAI track how the model is being used and filter outputs they deem potentially harmful.

Re: Imagen Video: high definition video generation with diffusion models

#444

Post by taylorius »

lolinder wrote: post #320
zakki wrote: post #316

Earlier quoted context omitted.

I don’t know why you put image and imagination in the same place. It is different. Imagination is the result of thinking. Image is the result of action. We can imagine killing people. Nobody harmed. But killing people in real life?

Image is not the result of action, it's the projection of imagination into the physical world. In your example, the correct parallel isn't killing people in real life, it's making an image of killing people. The ethics of making such images are debatable, but they already permeate our society without AI.

How do you know? The whole point of a deepfake is that it is indistinguishable from a photo of a real life event.

Re: Imagen Video: high definition video generation with diffusion models

#445

Post by chii »

derefr wrote: post #304
carlosdp wrote: post #300

Earlier quoted context omitted.

> There is a difference in ease of use. I could never use photoshop to fake something like that even if I wanted to. You couldn't, but basically any VFX shop easily could. Point is, it doesn't make anything possible that wasn't already possible, it just makes it more accessible. That's an inevitability with technology, as time goes on. The counter is not to try and suppress it, that has never worked and never will.

What motivation would a VFX shop have to do such a thing, though? (Money, sure, but what's a motive strong enough to be worth commissioning them?) It's always individuals who want to harm others in this particular way; and individuals don't throw around big-VFX-project amounts of money on petty revenge. But they'd certainly spend $20. DDoS attacks got a lot (1000x) more commonplace once there were DDoS services that…

and you can easily just drive your car into somebody to kill them.

DDoS services are almost always purely malicious (you can _may be_ argue that you can use them for pen-testing or load-testing). But cars are not just purely malicious; there's a lot of useful things cars can do, and that's why the dangers of car ownership is outweighed by the benefits, as judged by society - we just have some road rules, and licenses, so that people know to use them responsibly.

Why not the same with an AI model?

Re: Imagen Video: high definition video generation with diffusion models

#446

Post by chii »

TuringTest wrote: post #420
matheusmoreira wrote: post #346

Earlier quoted context omitted.

I can imagine a person killing other people. I can draw a person killing other people. I can make a computer draw a person killing other people. In all three cases, zero people were harmed.

But if you if you spread the image in a forum of haters, you may incite someone to kill that person. That's why there are laws against harassment. The danger is not the existence of an image, it is the act of communication that has consequences.

hate crime is already a crime. If someone were to want to incite hate, they could do it today just as easily as a hypothetical future where an AI generated image is readily available.

So the problem isn't the AI, it's the forum of haters. Restricting AI usage, in the hope of not having someone use it to incite hate is too roundabout a way to achieve any significant result, while everybody pays a high cost (of not being free to use such an AI as they see fit).

Re: Imagen Video: high definition video generation with diffusion models

#447

Post by BoxOfRain »

carlosdp wrote: post #285
atty wrote: post #266

Earlier quoted context omitted.

There is a clear risk from these sorts of models as they get better - I mean recreating specific individuals’ likenesses in compromising images (or even worse, video). We’re not at that point yet, but these things are getting better fast, so it’s only a matter of time. The problem is that there’s no way to mitigate those risks except to keep the model behind an inference-only API, or not release it at all - as soon a…

> There is a clear risk from these sorts of models as they get better - I mean recreating specific individuals’ likenesses in compromising images (or even worse, video). This has been possible without AI for a very very long time now (just open photoshop, etc). It barely ever happens, and society hasn't collapsed. I keep seeing this argument come up and it baffles me that informed technologists take it seriously, as…

People seem to forget that people as early as Joseph Stalin was all about manipulating images for unpleasant reasons, this isn't a new phenomenon at all.

Re: Imagen Video: high definition video generation with diffusion models

#448

Post by BoxOfRain »

TuringTest wrote: post #420
matheusmoreira wrote: post #346

Earlier quoted context omitted.

I can imagine a person killing other people. I can draw a person killing other people. I can make a computer draw a person killing other people. In all three cases, zero people were harmed.

But if you if you spread the image in a forum of haters, you may incite someone to kill that person. That's why there are laws against harassment. The danger is not the existence of an image, it is the act of communication that has consequences.

I quite like the concept of information hazards for reasoning about these kinds of risk.

Re: Imagen Video: high definition video generation with diffusion models

#449

Post by cheschire »

fny wrote: post #364
nperez wrote: post #318

Earlier quoted context omitted.

This thread is cathartic. I've been feeling uncomfortable with the level of control being sought over the usage of these tools for a while, but didn't want to ruffle the wrong feathers while just getting into AI as a hobby. I think there will be a pretty short window in which all of this hand-waving will be taken seriously. Not because AI won't be used for terrible things (I'm sure it already is) but because consumer…

> it doesn't matter how ethicists want "the general public" to use AI Something tells me these pricks will end up arguing for a reversion to thin-client compute. It is in their financial interest too after all.

7 hours after you posted this, the top article on HN is about how cloud desktops aren’t that great, and the comments in there are predominantly supportive of that sentiment.

Funny coincidence!

Re: Imagen Video: high definition video generation with diffusion models

#450

Post by stared »

i_like_apis wrote: post #261

The concern trolling and gatekeeping about social justice issues coming from the so-called "ethicists" in the AI peanut gallery has been utterly ridiculous. Google claims they don't want to release Imagen because it lacks what can only be called "latent space affirmative action". Stability or someone like it will valiantly release this technology, again and there will be absolutely no harm to anyone. Stop being so to…

It is not a matter of ethics. Do you think it is the priority for such corporations?

It is a matter of PR. It takes a single "problematic" generated content to be framed as "Google is sexist/racist/supports animal abuse", etc.

Statements related to ethics help in a few ways: holding secrets ("oh, we would love to share the models, but we cannot"), protecting against backslash (PR-wise, legal-wise), and PR on its own ("we are that ethical - see! it is even in our mission statement").

Post reply on HN