OpenAI collapses media reality with Sora AI video generator | If trusting video from anonymous sources on social media was a bad idea before, it's an even worse idea now
arstechnica.com
OpenAI collapses media reality with Sora AI video generator | If trusting video from anonymous sources on social media was a bad idea before, it's an even worse idea now::Hello, cultural singularity—soon, every video you see online could be completely fake.
You are viewing a single comment
Genuine question: why do we need this type of thing?
Especially in view of the harm it can cause, what's the point of creating this aside from generating shareholder value?
Sure, creating a video out of text quickly is cool, but is there an actual need for this?
I mean, the ability to generate whatever video you want without having to pay the costs normally associated with filming, location, actors etc is going to be very appealing to people like advertisers. This way you can have a few seconds of a beach for your travel company advert, for example, without having to pay for the stock footage or film it yourself. In fact I can see this transforming stock footage in general. Why bother to pay someone to make a generic video of 'people having a meeting' when an AI can do it for free in half the time. Doesn't even need to be that good if you're only using it briefly in a presentation. Not saying any of this is a good thing, but here we are..
I get what you're saying, but there isn't really any NEED for that
Thats kind of like saying we never needed cars because we had horses.
I mean, yes in so far that it opens those options up to people who may not have been able to afford it before. Whether that's a 'need' or not depends on your opinion of the company I guess.
Also there are other applications beyond this, of course. Easily made videos could help reduce the costs associated with treating some mental health issues for example.
The you have the ability for novice film makers to make content, a bit like how engines like Unity have made it easier for people to make games. Sure there'll be shit-tier tat, but there'll also be content made by creators that may never have had the chance otherwise.
It may help to make synthetic training data for other models / simulators
Which then just feeds to the system. But as this is a globally impactful thing, is there any real world need that outweighs the harm?
Cost savings are a need. It frees resources for everyone. Sure the vast majority of the profit goes to the shareholders but that's true of every labor saving device.
Do we NEED computers? You can hire people to do calculations by hand. The word Computer used to mean a job title, not a device.
OpenAI’s take is someone will create this technology - it might as well be them since their motivation is relatively pure. OpenAI is a non profit and they do work hard to minimise the damage their tech can cause. Which is why this video generation feature has not been launched yet.
OpenAI is no longer "pure." They are not open. They do not publish the details of any of the discoveries they've made (which used to be standard practice, even in the private sector). Their leadership is now in the "effective accelerationism" camp that worships capitalism, and sees developing AGI as their moral obligation, regardless of what harm it may cause to society. (They are also delusional, because it's very unlikely AGI will be developed anytime soon).
OpenAI is only technically non-profit. They're a proxy for Microsoft in all but name. They started out mostly pure, but their dickhead CEO has worked hard to undo all of that nonsense and has created parallel companies for OpenAI that can absolutely make profit while the main company gets to keep its nonprofit status. That was literally the entire basis for the board firing him (the CEO) a few months back.
Advertising holidays at places that do not exist! Exactly what we needed!
So no different than currently? Pictures and videos in ads all have heavy editing and post processing to make them look better.
It‘s highly debatable how much editing is ok or when it get‘s deceiving. Video can only show you so much of an actual experience, senses like smells, warmth and time need some editing to be somewhat communicated. That‘s just film making.
For an AI generated clip it‘s not even a debate. It doesn‘t exist and doesn‘t try to advertise anything. And since it‘s so much easier and cheaper than doing all of the above... Well let‘s just say the incentive to deceive was just pushed up to 11.
Uhh, ads already do that.
Usually their computer animations are a pretty obvious style choice. This is completely different than that and even more different than photos.
OK, poor example, but you get my point.
Art. Presentations. Visualizations. Porn.
Weird music videos. Just this week Youtube had pushed me music videos all done in the weird warpy AI style.
It's kind of cool and simultaneously already feels like a fad that tired with.
Who said anything about need?
It was created because someone thought of it. How it's used is a measure of the person using it.
People will find ways to utilize whatever someone creates. And usually in ways the creator never envisioned.
"Needing" something comes after a tool becomes ubiquitous. Imagine trying to screw in a Phillips screw with a slot screw driver - you'd need a Phillips driver because those screws are now ubiquitous (and I can't wait for them to go away. All hail our stripped screw saviour Torx!)