Writing is fundamentally the transfer of information from your brain to my brain. If you have 1000 bits of semantic information you want to transfer, you can't give 300 bits of semantic information to an LLM and have it fill in the remaining 700, because it doesn't know what those 700 bits are. If it's able to guess those 700 bits correctly, then they aren't true semantic information, and you really only have 300 bits you want to transfer. You might as well transfer those bits to me directly, rather than having the LLM add on an extra superfluous 700 bits that I then have to filter out.
No, thanks. Depending on the type of information being transferred and just how uncertain the recipient's knowledge is, I may or may not need to add some amount of information to whatever I have to share. That information would be extra context you may need to have, on top of what new thing I am trying to explain. If I know I'm talking with someone who's all caught up, great, no need for context setting. Though even in those cases, there might be some terminology I want to define, or to provide a list of acronyms or whatever. That is boilerplate work, but is at least useful, if not needed.
I know people tend to view this sort of comms as "technical" - they shouldn't be. So much misunderstanding happens because people are just not aligned on their shared context or medium.
Of course, AI is not fully reliable to generate that context, you as a writer need to put in the work to review and align the result with your own understanding, or drive it thoroughly before.
All that being said, I acknowledge that people (me included) love to be sloppy in their comms (with or without AI) and then blame others for misunderstanding. It's also unlikely we'll change soon. What can you do.
hatthew · · focus · HN ↗
Writing is fundamentally the transfer of information from your brain to my brain. If you have 1000 bits of semantic information you want to transfer, you can't give 300 bits of semantic information to an LLM and have it fill in the remaining 700, because it doesn't know what those 700 bits are. If it's able to guess those 700 bits correctly, then they aren't true semantic information, and you really only have 300 bits you want to transfer. You might as well transfer those bits to me directly, rather than having the LLM add on an extra superfluous 700 bits that I then have to filter out.
xenocratus · · focus · HN ↗
All that being said, I acknowledge that people (me included) love to be sloppy in their comms (with or without AI) and then blame others for misunderstanding. It's also unlikely we'll change soon. What can you do.