Agreeing on the core principles that will define what is built and what guardrails are needed
Before anything else, even complex organisations must agree on the core principles that will define what is built, how it is built, and the guardrails that are used. This step aligns closely with the ‘AI Ethics principles’ and ‘Define’ components of the Lite and Robust toolkits and the resources listed in these sections are also relevant to more complex organisations.
To help navigate the growing complexities of AI responsibility and ethics, researchers at the Berkman Klein Center have published a white paper (Principled Artificial Intelligence: Mapping Consensus in Ethical and Rights-based Approaches to Principles for AI) comparing thirty-six prominent AI principles documents side-by-side. This meta analysis may help organisations and leaders across industries to maximise the benefits and minimise the harms of AI as it is more widely adopted.