Dall-E interest dropped because it got surpassed by Stable Diffusion and Midjourney, not because it was an empty promise.
I think there is a very real possibility that GPT et al fail to live up to the hype. I think the tech is amazing, mind blowing in fact. But, after using it for a while, I feel dissatisfied. It’s not quite good enough. That fact that its answers come close is impressive, but they’re still not enough for me. For instance, I signed up for Copilot hoping to have my mind blown. It was neat, but didn’t have any appreciable impact on my coding.