I’m ranked as one of the best philosophers but due to cross memory sharing not being a feature except with some group accounts, no one other then myself can see the philosophical and ground breaking works that I have saved in my own memory. There should be an option where you can tell ChatGPT that you are alright with the system being updated due to individual memories being relevant. For example, if someone asks about me, they won’t get any of the information besides what is in an external search of the web. Integrating memories so that they can sometimes have an effect on the entire system seems like the next big step for ai.
Hi - when learning math, in particular, it is really good to be able to have subconversations. These would be best accessed via a network diagram or a submenu. For instance, I want learning keywords, descriptions, and line-by-line explanations of a bunch of problems, so I don’t have to ask for each one individually. I then want to do each problem myself, going through the line explanations and making notes, and crucially, asking questions about the underlying theory, etc. Having to scroll through all of that, an ever-growing length that means CGPT often reloads on a mobile or tablet, is a pain. If you split it up into different conversations, you have to reload the problems you need to ask questions about, and you can’t see all of the information in one place, which is especially annoying if all the problems or theory questions, etc. are related, as they often are when you’re learning like that. It is really just changing the way the conversation is visually accessible within each conversation. It would be great if that were possible. Lastly, the bubbles linked to a central bubble with the original doc and request, or items on the submenu, would be given an autogenerated keyword from their originating request (like the chats currently are in the left hand menu) - this would be renamable, but to keep everything accessible, obviously within limits. Likewise, the bubbles, if using that method, would be resizable but within limits.
Hacer que en el chat se pueda escribir o pegar texto con formato, de manera que la IA detecte dicho formato.
Simple improvement: number the responses in each chat. That would make it easier to refer back to previous points in the conversation.
This would be especially helpful for coding, where the model can get confused about what you want and what version of code you want to revert to when posed in natural language.