Project Brief Summary:
Artificial Intelligence has become a buzz word as of recent. With multiple large language models arising such as Meta's "Llama AI", Anthropic's "Claude", Open AI's "ChatGPT", and more, there is no shortage of tools. There are however problems with supporting these tools. MIT lists some of the concerns in their article "Explained: Generative AI's environmental impact". This article does not define all of the problems and positives of AI, however serves as a jumping off point for this experiment. The goals of this project is lofty with the intent to learn the device more than actually produce a fully functioning AI in a singular chip. The goal is to fail and see what pops out of the ashes.
The funder of this project understands that this is the goal of the project and is aware that the best case scenario they receive nothing on return but burnt development board. Requirements are designed to reflect a fully functioning project and may change with time and gaining of knowledge.
Tools:
GitHub Repository
Goals:
Create a functioning AI on a hybrid device
1. Have AI interact with user when only power is supplied to device
2. Have AI interact with user through keyboard input and switch input
Requirements:
Project Materials:
FPGA
Multimeter
Project Outcome:
Outcome of project is to host a large language model AI on a single chip.