OpenAI is making a huge mistake by deprecating DALL-E-3

windysoliloquy: > You guys know what I’m thinking of? What gpt-image-2 and multimodal learners can’t be: an unimaginable spectrum of keyword hallucinations evolved out of nothingness. Tiny but coherent (well..sort of). Where it also can’t succeed currently is in the artifact-free reality detailed deep in every pixel that others might provide.