Hi everyone! This is my first post here, and I’d like to share an idea.
Recently, I customized a CustomGPT that, instead of immediately providing an answer, first writes a visual, symbolic narrative for itself to delve deeper into the topic. It works well for me.
Now, I have a question: what if we used the Sora model to generate introspective imagery in response to a user’s prompt and then distilled insights from these imagined scenes into a verbal answer?
For example, at the end of May, we could ask GPT-5, “What is love?” Instead of answering outright, GPT-5 could first visualize various scenes related to love and draw inspiration from them. Isn’t this how we, as humans, often think?
Maybe silicon-based intelligence should have this “superpower” too.
I suppose it would be expensive, but on the other hand, the graphics in this imagination mode wouldn’t need to be highly detailed. When we, humans, imagine something, we don’t do it in ultra HD or with many overly complicated details. 
I would love to have this feature. What do you think?