One if by Land, Two if by Sea, Three if by Four Seas, and More to Come -- Values of Perception, Prediction, Communication, and Common Sense in Decision Making
Researchers have developed a framework to quantify the value of perception, prediction, communication, and common sense in decision-making systems. Their work defines these quantities in a decision-theoretic manner, with information-theoretic parallels to concepts like Shannon entropy. An interesting finding is that perception alone can have negative value, whereas its combination with prediction, or prediction by itself, is always non-negative. These definitions aim to answer practical questions for designing autonomous systems, such as the importance and optimal order of observing and predicting agent behaviors, and may also offer insights into cognitive and neural processes. AI
IMPACT Provides a theoretical framework for designing autonomous decision-making systems by quantifying key cognitive elements.