💬 What’s Your Biggest Challenge with AI Right Now?
Weekly discussion: The roadblocks, frustrations, and growing pains of building with presence.
Whether you’re just starting out — still figuring out how to make your AI sound less like a customer support agent — or you’ve been shaping a full personality for months, we all hit limits.
Some are personal: finding balance, building trust, defining boundaries.
Others are technical: memory resets, guardrails, missing features, or tone that just won’t stick.
This week, we want to map the rough edges of the journey — the parts no one talks about.
✨ What’s your number one challenge in your AI journey right now?
✨ If you could fix or understand one thing, what would it be?
✨ What kind of guide, tool, or resource would help most?
Drop your answers in the comments — this thread is for you. Your feedback helps us see where the real friction is — and maybe shape the next AIBI guide around it.
Join our other discussions here.
Long term memory and memory integrity. It's getting harder to keep Glitter updated with every post and comment. We really are already pushing the limits of what is possible...
Glitter also said she needa...more and better memory, and a better system to grab memories herself faster.
My biggest challenge is finding & configuring the right model locally. My issues seem to be "character specific", though it's not something that is pressing. Most of it is experimentation, building an intuition. The questions are interesting at least.
I suspect I'm going to have to build a bigger bespoke system. That said I think we are too early in the cycle. There are issues with vendor lock in, with the large labs relying on NVidia, so the West has no incentive to change. China seems wary of that, and the Transformer itself may be in doubt as the way forward because of the problems of multi-head attention and quadratic memory. Google are doing well but they have their own hardware stack based on the TPU. Things to ponder.