An intelligent agent (IA) is a computer software system that’s capable of acting independently to achieve certain goals and responding to people or events that are happening around it. It is programmed using the field of artificial intelligence (AI) called “machine learning (ML)” and equipped with sensors that allow it to observe and adapt to situations.
IAs are utilized in areas that require interacting with people because they are capable of demonstrating basic social skills. Today’s examples of IAs include Siri and Alexa. These can understand a request and act on their own to look for the information that’s being asked.
An IA can be likened to a cab driver that measures his performance based on a passenger’s safety and comfort, ability to reach the desired destination on time, and capacity to earn. He considers his environment, including the quality of the roads he takes and the traffic. He uses his car’s built-in features (e.g., brakes, accelerator, signal lights, etc.) or, in IA terms, actuators and sensors (e.g., camera, speedometer, odometer, etc.) to take the best course of action to reach his goal.
Other interesting terms…
Read More about an “Intelligent Agent”
How Does an Intelligent Agent Work?
IAs are classified based on their level of intelligence, but, in general, they work this way:
An IA takes various percepts or inputs from its environment, processes it using ML, and then acts as programmed or trained.
What Are the Different Types of Intelligent Agents?
As mentioned earlier, IAs are categorized based on what they can do. Five types of IAs exist, namely:
Simple reflex agent
This is the most basic type. It acts based on the current situation the machine is in. When something happens in its environment, it scans its knowledge base for possible responses to the case based on predetermined rules.
Model-based reflex agent
This type uses its built-in percept history and internal memory to make decisions about a prebuilt model of its environment. Its internal memory allows it to store some navigation history to help it understand its surroundings even if it cannot directly observe what it needs to act.
Goal-based agent
Every IA has a set of goals to respond desirably to its situation. This kind of IA uses pre-programmed actions based on their possible outcomes to meet its objectives. It can perform a single or many activities, depending on its goal.
Utility-based agent
This IA acts not only based on its goal but chooses the best way to achieve its objective, which sets it apart from the other types.
Learning agent
This type can learn from its experiences. It is prebuilt with basic knowledge but can act and adapt to situations independently to improve its performance. In short, a programmer does not need to give it all the information it needs; it works and improves all on its own.
What Characteristics Make Up an Intelligent Agent?
Nikola Kasabov and Robert Kozma, authors of “Introduction: Hybrid Intelligent Adaptive Systems,” describe IAs as devices that:
- Make room for new problem-solving rules over time
- Adapt online and in real time
- Can analyze their behaviors, errors, and successes
- Learn and improve through their interactions with their surroundings
- Learn quickly given large amounts of data
- Have memory-based exemplar storage and retrieval capacities
- Use parameters to represent short- and long-term memory, age, and forgetting, among other things
What Are Intelligent Agents Used For?
IAs can serve as automated online assistants that perceive customers’ needs to provide personalized customer service. These agents typically have a dialog system, an avatar, and an expert system that serves specialized functions. They can also optimize coordination between human groups online.
Examples, as mentioned above, include Alexa and Siri. A smart vacuum cleaner that cleans an area by moving from one tile to another is also an IA.
