Where is the line between similar content and spam in generative AI?
Hi everyone,
I've been contributing AI-generated assets to AS since August 2023 and my acceptance rate is around 85%, which I would guess is not terrible, although I don't have a lot of examples to compare it to. I am now in the process of uploading my next big batch of assets and I am curious about the spam policy as I tend to upload many similar assets.
I've read various articles on Adobe, including "Similar content versus spam" and I feel like I still don't have a good answer on this policy, at least in terms of AI-generated content. I've noticed that different AI-generated motifs are treated differently in the review process. For example, abstract assets may be accepted in bulk, even in batches of 20 or 50, but "realistic" or "photo-like" ones tend to get rejected for looking too similar. More confusingly, some abstract assets may be rejected in bulk, even though in their nature they are identical to those which were accepted, just with a different style, or flavor, if you will.
I do see value in uploading many similar assets, at least in the abstract category, as a customer may want to download multiple different variations of the asset they like, which has happened to me a few times, and was a pleasant surprise. However, I've read horror stories of contributors getting suspended or their accounts terminated citing this exact issue.
I am conflicted on this topic because, on the one hand, the review period for AI-generated assets takes up to two months, barely leaving any space for productive experimentation or safe trial and error, and on the other hand, I do see the value in similar assets in my portfolio, so I tend to upload many variations of the same prompt, which are hopefully different enough, to increase my portfolio size and earning potential.
I would love to hear your thoughts on this. Cheers
