Dots in ChatGPT muddles who you're actually talking to
The author says ChatGPT's Chat and Dot are two separate entities that don't share memory, unlike Meta Muse's single entity, and argues the main Dot should just be the app itself.
Dots in ChatGPT has confused my mental model of the app. I have long considered “Chat” or “Chatty” the AI I talk to. The entity. But now that entity has another entity living inside it. If you ask Chatty what you’ve been talking about with your Dot, it speaks about it in the third person. It doesn’t say “we were talking about X.” It will say something like “You and <dot name> we’re talking about X.” At least, that’s what I got this morning when I asked if it could see today’s conversations with my Dot. This breaks my mental model of ChatGPT. With Muse, for example, there is one entity. It is the app, AI, and agent. I’m always just talking to my Muse. It’s very clear in my mind. But this isn’t the case in ChatGPT right now, and I find it another thing that makes the app cumbersome compared to Dots. Everywhere I go in Muse, main chat or side chat, it’s always a single entity. But now with Dots it’s unclear who I’m talking to and what they remember. Chatty tells me it can see some of what I discussed with my Dot, but not everything. Do they share a consistent memory? Personality? Should they. This seems like a small thing, but it’s another point of friction. I think the main Dot in everyone’s account should just be a proactive version of the entity they’re already interacting with. Don’t put one entity inside another. Just make the main Dot the app itself or a representation of it. Then, sure, if people want to spin up a lot of Dots that aren’t the main entity, each of those can be separate and standalone. But the main Dot should just be the primary app entity itself.