What is an Agent?
- An agent perceives its environment and acts accordingly
- Reasons through information
- Takes actions to achieve a set goal
- Two main parts:
- Sensors -> gather information about environment
- Actuators -> perform actions
Read: Types of AI Environments
Types of Agents
-
This is our basic model of an agent:
From this basic model, we have four types of agents:
- Simple Reflex Agents
- Model-based Agents
- Goal-based Agents
- Utility-based Agents
Each agent is more complex than the last
Simple Reflex Agents
-
Visual model:
-
Makes decisions based “Condition-Action” rules
- e.g. if red traffic light, then stop
-
No memory
Model-based Agents
-
Visual model:
-
Stores memory of the past as an internal state
- Much better for partially observable environments
Goal-based Agents
-
Visual model:
-
Goals-based agents shifts the focus from immediate inputs to future outcomes
- Needs to reason and plan through multiple steps
Utility-based Agents
-
Visual model:
-
Cares not only about goal, but how well they reach it
- Can score their performance