AI will serve as a specialized tool under the complete control of trained practitioners, in accordance with our ethical AI usage principles and subject to oversight by a diverse panel of our peers. Its goals, scope, and limitations are defined exclusively by the humans to whom it is accountable. All outputs and actions of the AI system must be explainable, reversible, and directly attributable to operator intent.
Practitioners must determine the AI's functions and limits. The AI system should be under the complete authority of trained practitioners who define its functionality and limitations.
Every action taken by AI must be attributable to a human operator's intent. Accountability requires that every action taken by the AI system can be directly traced back to the intentions of a human operator.
Every action or decision made by the AI must be explainable in a way that a human staff member can understand and articulate. This principle ensures that human staff remain integral to the AI's operation.
Practitioners must always retain the ability to undo or modify AI actions, preventing full autonomy.
All AI usage must align with the organization's ethical principles.
All AI use should be reviewed by a diverse panel of peers.
In this context, AI refers to computer systems designed to perform tasks that traditionally require human intelligence. These tasks include problem-solving, perception, natural language understanding, and learning.