AI summary is based upon code including comments - may be unduly biased by comments

[RobertASmith]

Comments included in the code are likely to trump the code itself. There is no guarantee the code does what the comments say.

Options:

1. Disclaimer - "this has been fed to a LLM with comments and therefore comments may influence description"
2. Remove comments before passing to LLM.
3. Pass comments separately and tell LLM that it should say if it believes there is a difference between comments and code.