Challenges of conversations and not remembering user interactions

I want to discuss what I see as an architectural/scalability challenge and an efficacy challenge within ChatGPT. ChatGPT doesn’t carry information about previous interaction with a user from conversation to conversation. So ChatGPT isn’t developing profiles about a user that enable it to learn and work with individuals better over time. ChatGPT only seems to keep things in mind during a specific conversation. A user can work around this (hack this problem), in a way by keeping all of their interactions with ChatGPT within a single conversation or a few topical conversations. If individual conversations are used repeatedly, they get very large and I suppose ChatGPT would have to keep a lot in memory and this would be less efficient; i.e. scaling challenges. This could be remedied by having user profiles. Are my speculations on how this work correct? Is OpenAI considering allowing customers to opt-in to profile memory or something like that in the future?