As we have read in the previous articles that an agent is anything which perceives through its environment with the help of sensors and acts upon it accordingly with the help of actuators. For example, human beings, animals, robots and softbots all are agents. Now let us study what is an AI rational agent.
What is AI Rational Agent?
If an agent is able to take good decisions considering all the past as well as the current percept then the agent is said to be a rational agent. In other words, we can say that a rational agent is an agent which has the capability of doing the right thing at the right time.
Before going into details about rational agents let us first see what rationality actually means.
The meaning of Rationality
For anything to be termed as rational at a given point of time depends on the following things:
PEAS
Form the above figure; we can conclude that PEAS is a short form which is written for Performance, Environment, Actuators and Sensors.
- Performance Measure – It is a measure which defines the success or failure of an agent. Performance of each agent will vary with respect to its precept.
- Environment – It is the surroundings from where the agent will learn and react with the help of sensors and actuators respectively. Different types of environments (explained in the previous article) are faced by the agent in case it is set in motion.
- Actuators – The part of the agent from which it executes its output are called actuators.
- Sensors – The parts of the agent which help an agent to collect information regarding the environment are known as sensors.
Ideal Rational Agent
The above points lead us to the definition of an ideal rational agent. An ideal rational agent is an agent which can take various actions that would maximise the performance measure based on the perceptual history and the inbuilt knowledge of the agent.
Rational Agent – Functionality
Before an agent is actually put in the environment, the precept sequence and actions for corresponding precepts need to be fed into the agent. This helps the agent to start functioning with the help of basic inputs. Based on these inputs, the agent does its basic functions and keeps on learning about the environment. This increases the complexity of the agent’s learning.
For instance, if we consider the case of a vacuum cleaner, based on the amount of dirt in a room, it will decide over the amount of power needed to clean the room. The agent will learn this from the environment itself as this was not fed in the agent before.
So, an agent constantly learns from the environment and it upgrades and changes its perceptual experience. This is usually done with the help of learning techniques such as reinforcement learning.
Hence, performing required actions so that future percept can be modified is one of the main and important parts of rationality and it purely depends on the amount of exploration the agent does.
AI Rational Agent – Examples
Now, let us have a look at some of the examples of AI rational agents.
Example 1:
| Agent | – | Self-Driven Car |
| Performance Measure | – | Comfort, Safety, Time Taken, Correct Navigation. |
| Environment | – | Roads, Signals, other Vehicles, weather and Pedestrians. |
| Actuators | – | Steering Wheel, Brake, Horn, Accelerator, Indicators etc. |
| Sensors | – | Cameras attached, Speedometer of the car, GPS, Odometer etc. |
Example 2:
| Agent | – | Vacuum Cleaner |
| Performance Measure | – | Cleanliness, Battery life, ease of use, efficiency. |
| Environment | – | Room, floor, furniture, carpets, other objects. |
| Actuators | – | Wheels, brushes, vacuum extractor. |
| Sensors | – | Cameras, bump sensor, wall sensor etc. |
Example 3:
| Agent | – | Diagnostic System in Hospital |
| Performance Measure | – | Patient Health record, input costs. |
| Environment | – | Hospital, staff, patients. |
| Actuators | – | Diagnostic information, treatments, referrals etc. |
| Sensors | – | Keyboard (for entering data), patient’s replies, test reports etc. |
Example 4:
| Agent | – | Online Robot Tutor |
| Performance Measure | – | Student’s score in test. |
| Environment | – | Class of students, Classroom, desk, chairs, board. |
| Actuators | – | Display Screen, Suggestions, exercises, corrections etc. |
| Sensors | – | Keyboard, voice input etc. |
Rational Behaviour
For an agent to be a rational agent, it must have a rational behaviour as well.In order to decide if the behaviour is rational or not some penalties are applied on the actions taken by the agent when its performance is being measured.
Take for instance, in automated cars, success or failure would be measured on the basis of the following:
| ACTIONS | POINTS |
| Wrong Turn Taken | -10 |
| Breaking Traffic Rules | -5 |
| Reached Destination | 30 |
Now, the performance of the agent will be measured according to the above points. If the car reaches its destination safely then it will be awarded 30 points. But if it is breaking any traffic rules then 5 points (each for every rule break) will get deducted. And if the car takes any wrong turn, then 10 points will get deducted.
This is an example of penalty points. In a similar fashion there can be many other penalties and award points that can be levied on the agents.
So, we can say that an agent which will attain its goal with least number of penalties and has the maximum number of points will be termed as a rational agent. Hence, the basic idea is to optimize an agent so that it attains it’s goals on the basis of success criteria and the penalties which are associated with it in order to maximize its performance.
Final Words
We should note here that a rational agent is not the same as omniscience. This is because an omniscient agent actually knows what would be the result of its actions and it can act accordingly. This is not in case of a rational agent.
Finally, we can say that an agent is usually built with intent to satisfy the world of AI with its immense computing abilities and significant decision making. With enhanced learning and better coordinated activities better and more intelligent agents would be made.