If you craft the wording of the prompt just right, is there a way for Copilot to extract text from an image verbatim? (No reinterpretation; no missing words – just exactly as it reads?)

We’re having mixed success with this. Less complex images with minimal text seems to work pretty well. But once the text becomes lengthier and more complex, Copilot changes the wording. 

Is that simply the nature of AI that it will always do that, or is there a way to word the prompt (maybe over the duration of a conversation) that would capture it exactly? 

​If you craft the wording of the prompt just right, is there a way for Copilot to extract text from an image verbatim? (No reinterpretation; no missing words – just exactly as it reads?)
 
We’re having mixed success with this. Less complex images with minimal text seems to work pretty well. But once the text becomes lengthier and more complex, Copilot changes the wording. 
 
Is that simply the nature of AI that it will always do that, or is there a way to word the prompt (maybe over the duration of a conversation) that would capture it exactly?   Read More

​