Within the BOM I was hoping to have the following covered in a machine readable format, settling on a custom like Bibtex implementation. Easy enough for someone to read but also parsable by software if needed.
- Transparency: Providing clarity on the tools, hardware, data sources, and methodologies used in the development of AI systems.
- Reproducibility: Offering enough information for researchers and developers to reproduce the models and results.
- Accountability: Ensuring creators and users of AI systems are aware of their origins, components, and performance metrics.
- Ethical and Responsible AI: Encouraging the documentation of training data sources, including any synthetic data used, to ensure there's knowledge about potential biases, limitations, or ethical considerations.
I wanted to mention it here as with the realm of finance I think it's something to discuss. In preparing for my talk Data is Business, Business is Data: The 2023 AI Redux, which is this Saturday at the Northern Ireland Developers Conference (https://nidevconf.com). I was experimenting with some basic multiple linear regression and finding 0.2% differences in accuracy scores depending if I used R or Python.
As more and more models are created (look how many Web3 experts became AI experts all of a sudden!), it's becoming more and more important to trace lineage back to the model development. If the EU pass any AI regulation then this kind of thing will be important going forward.
The original article is here:
https://redmonk.com/jgovernor/2023/10/1 ... materials/
The core repository of the AI Bill of Materials:
https://github.com/jasebell/ai-bill-of-materials
