Four Principles of Explainable Artificial Intelligence

Abstract

All comments will be made public and are subject to release under the Freedom of Information Act (FOIA). Additional information on submitting comments can be found at https://www.nist.gov/topics/artificial-intelligence/ai-foundational-research-explainability. Trademark Information All trademarks and registered trademarks belong to their respective organizations. Call for Patent Claims This public review includes a call for information on essential patent claims (claims whose use would be required for compliance with the guidance or requirements in this Information Technology Laboratory (ITL) draft publication). Such guidance and/or requirements may be directly stated in this ITL Publication or by reference to another publication. This call also includes disclosure, where known, of the existence of pending U.S. or foreign patent applications relating to this ITL draft publication and of any relevant unexpired U.S. or foreign patents. ITL may require from the patent holder, or a party authorized to make assurances on its behalf, in written or electronic form, either: a) assurance in the form of a general disclaimer to the effect that such party does not hold and does not currently intend holding any essential patent claim(s); or b) assurance that a license to such essential patent claim(s) will be made available to applicants desiring to utilize the license for the purpose of complying with the guidance or requirements in this ITL draft publication either: i. under reasonable terms and conditions that are demonstrably free of any unfair discrimination; or ii. without compensation and under reasonable terms and conditions that are demonstrably free of any unfair discrimination. Such assurance shall indicate that the patent holder (or third party authorized to make assurances on its behalf) will include in any documents transferring ownership of patents subject to the assurance, provisions sufficient to ensure that the commitments in the assurance are binding on the transferee, and that the transferee will similarly include appropriate provisions in the event of future transfers with the goal of binding each successor-in-interest. The assurance shall also indicate that it is intended to be binding on successors-ininterest regardless of whether such provisions are included in the relevant transfer documents.

Explainable artificial intelligence: an analytical review

“…In recognition of the growing importance of this topic, NIST published in August 2020 Four principles of XAI (Phillips et al, 2020), which define the following fundamental principles which an AI must honor to be considered an XAI as follows:…”

A machine-learning approach to human footprint index estimation with applications to sustainable development

“…Furthermore, the use of innovative tools such as Layerwise Relevance Propagation indicates that the machine learning-based approach is correctly identifying human-made features and activities (e.g., transportation networks or land-cover conversion) providing confidence in, and interpretability of, the CNN’s predictions. This step is especially important if the ml-HFI is to be used for policy decisions as national and international laws have begun to require explainable AI systems for decision-making 41,42 .…”

Introduction to Interpretability and Explainability

“…2. Trustworthiness: According to NIST [Phi+20], the trustworthiness of an AI application is ultimately derived by its explainability. We attribute greater trust to AI algorithms that are relevant, easy to understand, and not prone to misrepresentation.…”