Read and Download Free eBooks
/
Categories
Browse by category
1,965 books to get lost inYours to read. Yours to keep.
No books found.
Try another title or author.
Try another title or author.
This NIST report proposes four principles for explainable artificial intelligence: providing an explanation, making it meaningful to its audience, ensuring that it accurately reflects the system's process and recognising the limits of the system's knowledge.
The authors relate these principles to different kinds of explanation and to the ways people interpret decisions. The discussion draws on work in computer science and psychology and considers the challenges of evaluating explanations. This complete 2021 NISTIR 8312 retains its examples, illustrations and bibliography.