Include Message Timestamps in LLM Context

Problem

ChatGPT currently lacks awareness of the elapsed time between messages in a conversation.

For example, I told ChatGPT that I was about to drive home. A few seconds later, it asked whether I had already arrived.

The model understands temporal relationships perfectly well but cannot apply that reasoning without access to message timestamps.

Proposed Solution

Include message timestamps (created_at) in the conversation context provided to the model.

This would allow ChatGPT to:

  • Distinguish between seconds, minutes, hours, and days between messages.

  • Better understand whether previously mentioned activities could have been completed.

  • Interpret follow-up messages more accurately.

  • Maintain more natural conversational continuity.

Why It Matters

Time is an essential dimension of conversational context.

The model doesn’t need a continuous sense of time. It simply needs access to temporal metadata that the application already possesses.

This seems like a relatively small architectural improvement with potentially significant benefits for conversational quality.

1 Like