salience

Terms from Artificial Intelligence: humans at the heart of algorithms

Page numbers are for draft copy at present; they will be replaced with correct numbers when final book is formatted. Chapter numbers are correct and will not change now.

Salience is the extent to which an event, object, or other stimulus captures attention. Typically things have high salience if they (i) stand out visually (or on other senory ways) from their backgorund; (ii) if they are less frequently experienced; or (ii) if they have greater potential impact. For example, (i) a bright red flower in the middle of a green field; (ii) an ice cream in the middle of winter; or (iii) a lion prowling. In fact a lion prowling in a shopping street is an example of both (ii) and (iii)! Salince is crucial to help humans and other animals choose were to focus attention and hence ac and learn more effectively. Computational models of attention have been particularly important in the development of large language models.

Used on Chap. 3: page 47; Chap. 5: page 103; Chap. 22: pages 539, 540, 546

Also known as salient