Author(s): Mehdi Zare

Originally published on Towards AI.

Building an AI chatbot that can hold a meaningful conversation over time isn’t just about choosing a powerful language model — it’s also about giving it a memory. As a data scientist transitioning into AI, I quickly realized how crucial it is for a chatbot or agent to remember context. In this overview, I’ll discuss the concepts of short-term vs. long-term memory in AI conversations and then compare several popular open-source tools that add memory to chatbots. My goal is to give you a general idea of how these tools work and when to use each one, without getting into individual implementation details.

Short-Term Memory (within a conversation session) refers to the immediate context the AI can recall from recent conversations. The model’s context window limits the amount of text that can be included in prompts. For example, GPT-4 might allow up to 32k tokens (roughly 50 pages of text) in its context. Everything the model “remembers” has to fit in that window.

If the conversation goes on for too long, early messages will eventually fall out of the context window and be forgotten. Many chat implementations, by default, include the entire chat history with each new iteration until they reach the… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI