| Summary: | We present an update to our multi-agent adaptive learning framework. The original prototype presented an adaptive learning framework using multiple large language model agents using OpenAI's GPT-4o model. Each agent is specialized to a specific aspect of adaptive learning. The agents communicate with each other using the autogen multi-agent framework. This paper extends our previous work by providing cloud-based account management, bookmarks, progress, and leaderboards. It also expands the subject area beyond mathematics and includes a Telugu language learning module. Other improvements include enhanced state machines, speech input, real-time chat capabilities, and encrypted progress storage. The code implementing the experiments is open-source and available on github1.
|
|---|