-
Notifications
You must be signed in to change notification settings - Fork 3
Open
Description
The note does not currently mention a number of inherent ethical aspects of using AI, including:
- major sustainability impact of both training and prompting language models, including enormous greenhouse gas emissions, water usage and raw material usage
- copyright violations by major AI providers that is inherent to the technology (eg Open AI told the UK House of Lords they cannot make AI without violating copyright, paraphrasing)
- likelihood of the propogation of bias and hate towards minorities, that is often not disclosed (or knowable by engineers building these models because of the size of their data sets). See for instance this paper on hate in multimodal datasets
My suggestion would be to add an ethics section to the note, that warns of the risks and recommends defaulting to not using these technologies given those ethical considerations.
That way, the note better aligns with the no harm, all people and sustainability principles in the Ethical Web Principles.
stevefaulkner, iadawn, aardrian, kfranqueiro, julianmka and 9 more
Metadata
Metadata
Assignees
Labels
No labels