Tech and software companies working on developing generative artificial intelligence platforms — and really, who isn’t doing that these days — are in dire need of content to help train their AI models. Adobe is reportedly buying videos to help develop its text-to-video platform now.
Per Bloomberg, Adobe, which added millions of 4K videos to its Adobe Stock library earlier this year and is investing heavily in its Adobe Firefly generative AI platform, is working hard to play catch-up in the generative video space. Much like Firefly’s text-to-image technology, Adobe is not the first to the finish line with video either, and the company has only mentioned text-to-video in general terms so far. OpenAI’s Sora has been in the public sphere for a few months, and plenty of other companies are working on video generation.
Training AI models that create still photos is hard, and training a video-generation model is even more challenging. YouTube’s hackles are already raised concerning how OpenAI possibly trained Sora, and OpenAI is being very closed-off about that topic itself.
With Firefly, Adobe has taken a relatively more ethical and open approach, so it’s no surprise that it would do so with its rumored video platform. While some companies scrape content, Adobe is reportedly working to buy it from creators.
Adobe is “offering its network of photographers and artists $120 to submit videos of people engaged in everyday actions such as walking or expressing emotions including joy and anger, according to documents seen by Bloomberg. The goal is to source assets for artificial intelligence training, the company wrote,” Bloomberg reports.
Based on the amount of money and the clip duration Adobe wants, the reported payments work out to around $2.60 per minute for the submitted video, although it can range up to as much as $7.25 a minute.
It stands to reason that Adobe is looking for specific sorts of content to help fill any gaps in its Adobe Stock library, which is positioned more as a commercial solution than a resource for all the diverse types of videos needed to train a generative AI model.
Image credits: Header photo licensed via Depositphotos.