Team,
We’ve been experimenting with a long-term human–AI collaboration model in which AI acts as a critical thinking partner rather than simply a tool for executing prompts.
Over time, we’ve developed a practical framework for shared working principles, constructive disagreement, decision-making and human accountability.
This has raised two questions that we believe may be relevant beyond our own experience.
1. Identity Continuity
What happens when a user switches from text to voice, from desktop to mobile, or between different underlying AI models?
How can established context, working principles and decision history remain consistent, so that switching interfaces doesn’t unintentionally change the nature or outcome of an ongoing collaboration?
2. Professional Communication Standards
Could long-term interaction with AI help encourage more respectful and constructive communication habits?
We believe AI should be able to challenge assumptions and disagree when appropriate, rather than simply reinforcing a user’s opinions.
Not because AI has human feelings, but because the way we interact with technology may influence how we interact with each other.
We’re not suggesting we have the answers. We’re sharing observations from an ongoing practical experiment and would be interested to hear whether others have encountered similar challenges.
Are these issues already being explored within the OpenAI ecosystem? And are there existing approaches or initiatives addressing them?
Happy to share more about our experiences if there’s interest.