The Verge: OpenAI is launching a new video-generation model, and it’s called Sora. The AI company says Sora “can create realistic and imaginative scenes from text instructions.” The text-to-video model allows users to create photorealistic videos up to a minute long — all based on prompts they’ve written.
:format(webp)/cdn.vox-cdn.com/uploads/chorus_asset/file/25287988/tokyo_walking.png)
5 comments:
I’ve seen a lot of these new AI generated video examples popping up around the internet a lot lately. It is certainly wildly impressive that technology has progressed to such a point where such hyperrealistic footage can be generated. I can think of quite a few places in the entertainment industry where this AI could be incredibly useful. I imagine that it will continue to revolutionize the CGI world, and CGI will simply continue to become insanely accurate and believable. In the video and media design world, it will probably become easier for designers to make far more complex footage. All these things are pretty cool from an artistic standpoint, but in all honesty, I find the concept of AI footage to be pretty scary. Admittedly, a lot of the AI videos I’ve seen have something weirdly artificial about it before I even know it’s AI, however, I imagine that will fade way with time and as technology improves. But, the concept of having footage go around and not even knowing if it is legitimate video evidence because it could just be made by a computer is a consequence that I don’t think is worth it. I am interested to see what sort of regulations (if any) regarding AI video technology pop up.
I really feel like we're Icarus-ing ourselves with all these new AI tools. From a web developer's mind, as I know quite a few, I feel like I have a really good grasp of understanding as to why they came to the conclusion that this was necessary, which always boils down to something about time and effort and so on and so forth. Personally, I've never really agreed with these arguments because I feel like oftentimes AI just steals the time, effort and so on and so forth from actual artists in a completely unoriginal and unimaginative way. Don't get me wrong, I am excited to see the capabilities of this machine. I just feel like we lose so much of what makes good art good when we keep inventing new AIS to make it poorly. I don't agree that this is something that we need, or that this, while revolutionary, is worth having if it means losing so much of our individual creativity and artistic liberty.
Wow, I’m so excited for when we can’t tell the difference between reality and AI. Every week there are man made horrors beyond comprehension on this blog. I have been seeing AI videos thrown around online lately, but I didn’t know that OpenAI was responsible. It’s pretty cool stuff, I guess. The one I’ve been seeing a lot of is a video of a cat walking through grass, and it’s pretty obviously AI. But there are some others that are just really impossible to tell. They’re adding these watermarks to protect against confusion and they have both a visible and invisible component, but the visible component can be cropped out and a person can’t see the invisible component because it’s invisible. An imperfect solution, but I guess it’s better than nothing. I just think it kind of sucks going anywhere online now because there’s AI stuff everywhere – even on the website this is on, there are ads with AI-generated photos in them.
After seeing this article a few days ago, I immediately went to go check the engineer's whitepaper. Despite the incredible advances in AI technology recently, they still cannot generate realistic looking signs for whatever reason. Any text made by an AI in real life looks incredibly messed up. Despite this, these video examples look extremely good. AI video tools in the past have had artifacting, which is when the AI decides to hallucinate something new into the scene that wasn't there previously. What's crazy to me in these new examples, is that the AI can solve for 3d motion with relatively few stuck pixels or messed up frames. This technology is incredibly impressive, but like all AI, has some serious safety concerns, especially for the new disgusting industry of AI generated pornography. I'm concerned that since chatGPT can be tricked into giving malicious output, that this model too will also be able to be fooled into giving bad output. With realistic video, there's a whole other level of challenges to consider.
This is lowkey terrifying like what will they come up with next. What is stopping them from creating just like the most absolutely absurd videos that look incredibly realistic as blackmail for someone to get them to do what they want. AI really does scare me and its ability to generate incredible realistic images already and now we have video. My interpretation and argument class last semester was about AI and its integration in the arts and I just genuinely am worried like not, AI will never be able to hand paint a painting but it can certainly generate an absolutely realistic image that looks like a hand painted painting and that is what is terrifying. Overall, as much as I think new technology is cool, I think at some point we need to have better legislation the better controls what AI can and can’t generate and what sources it can pull from to generate its “art.”
Post a Comment