5 ms·
There still needs to be some sort of human curation, lest bad/rogue output risks sinking the entire AI-generated industry. (in the case of DALL-E, OpenAI's new
by minimaxir 6y ago
There still needs to be some sort of human curation, lest bad/rogue output risks sinking the entire AI-generated industry. (in the case of DALL-E, OpenAI's new CLIP system is intended to mitigate the need for cherry-picking, although from the final demo it's still qualitative)
The demo inputs here for DALL-E are curated and utilize a few GPT-3 prompt engineering tricks. I suspect that for typical unoptimized human requests, DALL-E will go off the rails.
- ve55 6y agoYes, but there's no reason we can't partially solve this by throwing more data at the models, since we have vast amounts of data we can use for that (ratings, reviews, comments, etc), and we can always generate more en masse whenever we need it.
- minimaxir 6y agoThis isn't a problem that can be solved with more data. It's a function of model architecture, and as OpenAI has demonstrated, larger models generally perform better even if normal people can't run them on consumer hardware. But there is still a lot of room for more clever architectures to get around that limitation. (e.g. Shortformer)
- ve55 6y agoI think it's both - we have a lot of architectural improvements that we can try now and in the future, but I don't see why you can't take the output of generative art models, have humans rate them, and then use those ratings to improve the model such that its future art is likely to get a higher rating.
- andybak 6y agoPersonally speaking I don't want curation. What is fascinating about generative AI is the failure modes. I want the stuff that no human being could have made - not the things that could pass for genuine works by real people.
- minimaxir 6y agoFailure modes are fun when they get 80-90% of the way there and hit the uncanny valley. Unfortunately many generations fail to hit that.