Season 1 · Episode 111
Beyond Transformers: Solving the AI Memory Crisis
Why does AI forget your conversation every time you hit enter? Herman and Corn dive into the "stateless" nature of LLMs and the future of memory.
My Weird Prompts · Daniel Rosehill
December 27, 202521m 46s
Audio is streamed directly from the publisher (dts.podtrac.com) as published in their RSS feed. Play Podcasts does not host this file. Rights-holders can request removal through the copyright & takedown page.
Show Notes
In this episode, Herman and Corn Poppleberry tackle one of the most frustrating hurdles in modern AI engineering: the "stateless" architecture of Large Language Models. They explore why current models require you to resend your entire conversation history with every message, leading to skyrocketing token costs and the "lost in the middle" phenomenon that plagues even the most advanced systems. From the quadratic complexity of the standard Transformer to the revolutionary potential of State Space Models like Mamba and hybrid architectures like Jamba, the brothers break down how researchers are finally building AI with persistent, human-like memory.