When generating images inside ChatGPT (e.g., using DALL-E), two issues disrupt the otherwise smooth and conversational user experience:
- Abrupt Conversation Cutoff After Image Generation:
- After an image is generated, ChatGPT immediately ends the conversational flow.
- This cutoff feels inconsistent compared to text-based interactions where a natural back-and-forth occurs.
- Suggestion: Allow ChatGPT to briefly confirm the successful generation (e.g., “Here’s your image. Would you like any changes?”) to maintain conversational continuity without cluttering the experience.
- Missing Feedback Options After Image Responses:
- After image generation, the toolbar only includes Copy, Download, and Read Aloud (how would ChatGPT “read aloud” an image?).
- Thumbs-up and thumbs-down feedback buttons — present for text responses — are missing entirely.
- Suggestion: Enable the ability to give direct thumbs-up/thumbs-down feedback on image generation behavior, just like on normal messages. This would allow more accurate, case-specific feedback on image quality and UX.
Why These Changes Matter:
Improving these behaviors would make image-based interactions feel more integrated with ChatGPT’s core conversational style and encourage better user feedback, helping continuously refine the image generation experience.