Structured lists & Bing Bugs

I have stumbled upon a peculiar, albeit niche, issue while using ChatGPT in conjunction with Bing Browsing, and would appreciate if anyone could replicate it to ascertain its scope. My aim was to extract information from various pages of a website for personal consolidation. Anticipating a potential hitch in continuing the list numbering across different pages, I decided to track the numbers separately, planning to merge them later. However, I ran into a snag when the list didn’t commence at specific points as expected (e.g., it should have started with #16, but reset to 1). Resigned, I was ready to manually adjust the numbers, but to my astonishment, upon copying the reply, the full list was accurately copied to my clipboard. The visible output in chat displayed ‘1-7’, but on my clipboard, it rightly showed as ‘1-22’.

Adding to the complexity, when I instructed to ‘fix x in 7’ (presuming 7 should be 22 in this context), the edit was made to #7, which might just be an issue of semantic understanding, or perhaps something deeper. This bug has morphed my frustration into curiosity. My experience leads to questions whether this is an isolated occurrence or more widespread? Or perhaps, adjusts tinfoil hat, I’ve stumbled upon a part of the larger ILLUMINAI conspiracy? Idk ok I’m done

I think you describe markdown formatting, which will take precedence over the actual numbers:

  1. You start a numbered list
  2. you put another list item as three, but get “2” rendered.
  3. And by point 5, you see what’s going on.

Continuing, counting down today’s top 5 oddities:

  1. usage page shows no usage
  2. rate page shows no rates
  3. browsing doesn’t browse
  4. where’s code interpreter
  5. markdown works as designed.

Actual forum input: