1. Some current applications of computer vision available on our smartphones include facial recognition, document scanning apps (OCR) and AR filters like on Snapchat.
(ref)
2.
1. I took the picture, then ChatGPT processed the image, adjusting lighting, detecting edges, etc., then it detected boxes within the image, classified each box with the help of its training data, and output the results.
2. Some objects look similar to other objects so the AI identified it wrongly, like mistaking a flat round brush for a toy, and some items are blocked on the picture.
3. To draw on the photo, GPT had to regenerate it, which distorted the photo somewhat.
3. I don't trust facial recognition used in surveillances. Look-alikes are out there, and a model could misidentify a bad actor.
(ref)
I didn't use any references for the questions in part 2 as most are from your own observation.