AI Agents Overview and Types

Study Notes

AI systems focus on rational agents that sense and act within their environments.
Agents utilize sensors for observation and actuators for action.

Human-Agent: Uses sensory organs (eyes, ears) and actuators (hands, legs).
Robotic Agent: Employs cameras, infrared sensors, and motors.
Software Agent: Operates through keystrokes and file contents, displays output through screens.

Sensors: Detect environmental changes, providing information to electronic devices.
Actuators: Convert energy into motion, responsible for system movement (e.g., electric motors).
Effectors: Physical devices that influence the environment (e.g., wheels, legs).

Employs climate control algorithms to optimize comfort while minimizing energy consumption.
Balances conflicting goals and weighs the likelihood of success against goal importance.

User Input (Sensor): Collects commands via microphone or text.
Internal Model (User Preferences & Context): Tracks preferences for personalized interactions.
Learning Element: Utilizes machine learning to enhance capabilities.
Performance Element: Executes tasks based on the current state.
Critic: Evaluates agent actions through user satisfaction.

Reflex Agents: Respond immediately to stimuli without considering past states; rely on the event-condition-action rule.
- Real-Time Example: Automatic Light Switch, which activates with motion detection and deactivates when no motion is sensed.
- Limitations: Very basic intelligence and lack of knowledge about non-perceptual state aspects.
Model-based Agents: Utilize a more extensive view of the environment to inform their actions unlike reflex agents.