Hi everyone,
I’d like to propose a feature for OpenAI models to track the time elapsed between user messages, which could significantly improve context awareness in ongoing interactions.
The Problem:
When using a previous conversation as context (e.g., while coding in Python or Mathematica), the model treats older and newer messages with equal importance, regardless of how much time has passed. This can result in responses that may not fully reflect the user’s immediate needs or current focus.
Proposed Solution:
Incorporate time awareness to prioritize newer messages appropriately based on the length of the gap. For example:
- If there’s a long break, the model could respond with more emphasis on the most recent message, assuming previous issues have been resolved and treating it as a new task within the same framework.
- For short breaks, it should seamlessly continue the conversation, maintaining continuity and focusing on solving the current problem without treating it as a new request.
Example:
I’m using the same chat to write Python code with specific commenting and formatting requirements. At the beginning of the conversation, I explained these requirements (e.g., functions should have docstrings, and variables should follow snake_case naming).
- Week-long break: Chat should assume that previous problems are resolved, and I am giving a new task within the same framework. The response should focus on the new task, applying the previously established requirements without revisiting or continuing old discussions.
- Minute-long break: Chat should assume we are still working on the same task. The response should focus on solving the current problem, maintaining continuity, and not treating it as a new or independent request.
Benefits:
- Improves handling of long and short pauses in conversations.
- Ensures that ongoing interactions adapt naturally to user intent without unnecessary assumptions.
- Enhances overall fluidity and makes revisiting old threads more efficient.
This feature would make interactions more practical and focused, especially for tasks involving ongoing projects or technical discussions.
Does anyone else feel this feature would improve their workflow? I’d love to hear your perspectives or additional suggestions on this idea!
Best regards,
Misha