Hey open-source hardware makers! It might be time to update the Open Hardware Definition, it’s over 10 years old (here is my post on MAKE Magazine from February 10th, 2011). A lot has changed in the last 10+ years for open-source hardware and open-source software, and some things have not! There was/is an Open Source Hardware (OSHW) Definition 1.1 draft, but has not been updated on the wiki since December 10, 2018.

While there are a few things I’d update on the Open Hardware Definition 1.0 my focus is on adding something to address AI/ChatGPT/humans working with LLMs, etc. I’m going to propose the same thing that I’ve attempted to have the Open Source Initiative consider for their OSI + AI license / definition. Here’s a blog post about that as well. The goal is sharing which exact tools were used and in what ways to allow others to replicate (and iterate) with AI/LLMs, etc. it’s a little different than commenting code, or publishing code under an open-source license, but the intent can be the same.

My addition to the definition is specific to this “freedom”

“Study how the system works and inspect its components.”

The OSI + AI definition at this time leaves out the inspection of prompts and data access transparency, so here’s the proposed addition to the OSI + AI definition, and an update Open Hardware Definition or a parallel definition that is specific to AI.

Inspection of Prompts and Data Access Transparency:

In addition to the existing requirements, the preferred form for making modifications to a machine-learning system shall include access to the prompts and commands used during the training phase and/or code and hardware creation. This will enable users to understand the context in which the model was developed, including:

• Prompt Transparency: Access to a detailed log of all prompts, commands, and instructions used during the training phase and/or code and hardware creation, ensuring that users can see the exact inputs that shaped the model’s behavior.
• Justification and Documentation: Each prompt should be accompanied by documentation explaining its purpose, how it was constructed, and its expected impact on the model’s development.
• Replicability and Testing: The framework should provide means for users to replicate prompt scenarios to test modifications and understand their effects on the model’s outputs.
• Prompt and Model Linking: Direct links to the specific model versions used along with the corresponding prompts, enabling a traceable lineage from input to model behavior.
• Timestamp and Metadata Documentation: Each entry of the prompt log should be timestamped and include metadata such as the version of the model used at that time.
• Public Access to Logs: Where possible, logs of the prompts should be made publicly available, with links provided in the documentation to ensure that users can review the historical context and development trajectory of the model.

This addition aims to enhance transparency and foster an environment where users can more effectively audit, replicate, and modify AI behavior.

And of course we have a real-world example, we’ve been doing this for about 1 year! Check out our video, and article “Writing an Arduino driver with OpenAI ChatGPT and PDF parsing” and here’s an example of the prompt transparency when publishing open-source code.

What’s next? Proving that there is enough demand within the open-source hardware community to actually try and update the Open Hardware Definition for AI , either as a revision of the existing definition or a parallel one for AI. If there is, we can figure out what a legitimate process would look like and how it would work. I’ll email OSHWA, hit the forums / various Discord(s), and email open-source hardware makers.

If there isn’t any interest in an update, I will probably publish Open Hardware Definition for AI which would be the current definition, with the AI additions and since AI also stand for Adafruit Industries, I suppose it would be what Adafruit uses when Adafruit refers to open-source hardware and AI was used in some way transparently and others can adopt it over time (or not). I guess we’d need a logo too.

Related:

• There is a new draft of the Open Source AI Definition from the Open Source Initiative.
• What to Do When the Ghost in the Machine Is You Limor Fried’s code is often a de facto standard, and now ChatGPT is using it.
• ChatGPT Creates Arduino Drivers in the Style of Adafruit’s Ladyada.
• Putting AI in the Driver’s Seat – Adafruit experiments with AI.
• Anthropic publishes the ‘system prompts’ that make Claude tick.
• The Open Source logo(s).
• 20 years of open logos and gear logos.

*header image background question marks made with DALL-E 2 and GPT-4o Aug 29, 2024.