AI generated component should generate component wisely.

"wisely" and "AI" in the same sentence 🤔🤣

Here's the scenario that I tested. You can see that in the before video it uses useState hook and after changing the prompt its evident that in the after video it does not include the useState for the same prompt, even though it does not generate the exact same component(which is fine as we can't expect AI do that always).
Moving this to NR.

Re #4 : I think it can act a little bit wisely if we instruct it to perfection 😂.

What is the criteria for determining that this is fixed given the non-probabilistic output? e.g. what percentage of attempts before the MR generated extraneous code, and what percentage afterwards? The before and after videos show one attempt each, but that could be within the normal deviation of either code base.

Re #7:- It was failing/generating wrong component almost 6-7/10 times before this fix and it now does not generate wrong component or use unnecessary code in the component. I tested it 7-8 times but it always generate the optimised code using the same prompt as used in the videos in #5 📌 AI generated component should generate component wisely. Active .

Tested this MR 4–5 times using the following two prompts:
I’d like to create a CTA component which includes fields for heading, text, button and image.
Create a counter component with "+" and "–" buttons. When a user clicks "+", the number increases by 1, and when they click "–", the number decreases by 1. The current count should be displayed between the buttons.
useState was used every time in the second prompt (counter) and never in the first (CTA), as expected.
Moving it to RTBC.

@catch, our target success rate is 5/5 or 9/10.

Merged and backported

Automatically closed - issue fixed for 2 weeks with no activity.