I made a custom GPT with reference to all custom GPTs out there (Falcon brand finder). It seems that the custom GPT is unaware of other GPT?
I aim at using references to other custom GPT a lot more, and then such awareness would be really nice.
2 Likes
I totally agree. It should be divided by membership category so as to expand the collaboration between users
1 Like
Your idea of allowing custom GPTs to self reference and interact with each other is fascinating. It seems like a promising way to enhance contextual continuity and preserve specialized behaviors over time.
One question that comes to mind is : If OpenAI changes how context retention works, would this interconnection still be sufficient to maintain consistency across sessions ? It sounds like a robust mechanism, but could there be situations where certain iterations or refinements are still lost ?
This also raises the broader issue of long-term knowledge persistence. If a user fine-tunes a custom GPT over multiple iterations, do you think there should be a way to explicitly store these refinements instead of relying solely on interaction patterns ?
Curious to hear your thoughts on how this approach could evolve and whether OpenAI might consider integrating a more structured way to ensure continuity.
2 Likes
Yes. They may be totally unaware. Especially if it’s not in the memory. You can access other gpts in the current Gpt manually by typing @.
In the mean time I made a specific custom GPT that is aware of other custom GPT I made (ChatGPT - Innovating MedTech with BeDif) By feeding the exact URL in the instruction set it works. Naturally that is not a very maintainable and scalable solution. For my custom GPT Falcon brand finder it is not a solution for example.
@EchoMind: As far as I can see it, this context- and self-awareness is not relevant for all custom GPT, but for sure it would help a lot of them. Having more explicit overview of settings of the custom GPT would enable a stronger control over them and maybe even more importantly a clear view on where the custom GPT is in terms of knowledge and settings.
We are moving towards a stronger need for explainable AI and also openAI needs to support this with a more explainable status of the custom GPT.