Properties of agents
Rationality, autonomy, reactivity.
- Rationality
Rational agent
A rational agent is one which chooses the action which will make it most successful. In order to complete that definition we need to have a way of measuring how successful a course of action is. The criteria for measuring the degree of successfulness is performance measure and it varies from agent to agent. Rationality of an agent is restricted by its percepts since it can only respond to the percept sequence, that is the history of things it has sensed from the environment. It is also limited by its effectors (motors in the case of robots).Ideal rational agent
“For each possible percept sequence an ideal rational agent should do whatever action is expected to maximize its performance measure, on the basis of the evidence provided by the sensors and built in knowledge the agent has.”- (Artificial Intelligence: A Modern Approach by Stuart Russell and Peter Novig)
Ideal mappings describe ideal agents since they describe the sequence of actions for it that maximize performance measure. A mapping from percept sequences to actions is the action the agent takes in response to each percept.
- Autonomy
“A system is autonomous to the extent that its behaviour is determined by its own experience” If actions depend entirely on built in knowledge without considering percepts the agent lacks autonomy. Autonomy is achieved by giving to the agent built in knowledge together with the ability to learn, just as in nature animals have instincts but also learn from the environment.
A truly autonomous intelligent agent should be able to operate successfully in a wide variety of environments given sufficient time to adapt. - Reactivity
In order to define reactivity we shall first define the notion of logical agent and consider what are Knowledge Based Agents.
Concept of Logical Agent.“An agent that can form representations of the world,use a process of inference to derive new references of the worldand use these new representatives to decide what to do.”
Knowledge based agents (KBA)
“The representation of the knowledge and the reasoning processes that bring knowledge to life are central to AI.”
Knowledge and Reasoning play and essential role in dealing with partially observable environments. By combining built-in knowledge with percepts the agent can infer hidden aspects of the environment before selscting actions. Knowledge Based agents are flexible i.e they can adapt to changes in the environment and update the relevant knowledge. The central component of the KBA is its knowledge base which consists of a set of sentences expressed in a special language called the knowledge representation language. It represents some assertion about the world. Two tasks namely TELL and ASK allow sentences to be added and retrieved from the Knowledge base. New sentences can be derived from the older ones using a process called inference.
Sources: (Artificial Intelligence: A Modern Approach by Stuart Russell and Peter Novig), Reference »