Intelligent agent
An intelligent agent is a software program designed to make decisions, respond to its environment, and take actions to achieve specific goals, often employing artificial intelligence techniques. These agents operate autonomously, using sensors to gather data from their surroundings and actuators to deliver outputs. They can learn and adapt over time, improving their functionalities based on feedback. Intelligent agents range from simple digital assistants like Amazon's Alexa and Apple's Siri to more complex systems such as autonomous robots.
The evolution of intelligent agents traces back to early computational theories, with influential concepts such as the Turing Test introduced by Alan Turing in 1950 to evaluate machine intelligence. Intelligent agents are categorized into several types, including simple reflex agents that operate based on immediate conditions, model-based agents that incorporate historical data, and learning agents that refine their decision-making through experience. Practical applications of intelligent agents are widespread; for instance, they help manage non-human tasks in autonomous systems like robot vacuums and even fraud detection in financial systems. As technology advances, these agents continue to integrate more deeply into everyday life, enhancing user experiences and automating complex processes.
On this Page
Subject Terms
Intelligent agent
An intelligent agent is a software program that can make its own decisions, respond to its environment, and take action to achieve a goal. Over time, an intelligent agent can improve its function through the process of learning. The term is most often used in conjunction with artificial intelligence (AI).
All computer programs require some user input to set up the parameters of their operations. An intelligent agent takes those parameters, seeks additional input, adapts to the new input, solves problems, and moves toward accomplishing a specific goal. An intelligent agent can range from an AI-powered chess program capable of defeating a grand master to a “smart” thermostat programmed to adjust temperatures based on environmental changes. Virtual assistant technology, such as Amazon’s Alexa or Apple’s Siri, are also examples of intelligent agents.
Background
The earliest incarnation of computers dates back to the nineteenth century when some scientists developed computing machines that could compile data and make mathematical calculations. The first electronic computers debuted in the 1940s, but like their earlier counterparts, they could not “think.” They could only execute the commands given to them and could not store and retrieve information.
British mathematician Alan Turing theorized that it was possible to build a machine that could store, process, and retrieve information similar to the human mind. In 1950, Turing wrote a paper that laid out rules for determining if a computer was capable of human-like thought. His method, known as the Turing Test, had a subject ask a series of questions that were answered by both a computer and a person. If the computer was able to fool the subject into thinking it was the person, the machine was considered to have intelligence.
As computers evolved from the 1950s on, researchers developed programs that began to exhibit signs of AI. Checkers- and chess-playing programs were developed in the early 1950s and by 1966, the language-processing program ELIZA, the ancestor to today’s Alexa and Siri, was one of the first computers to pass the Turning Test.
In the 1970s, researchers believed that they were on the cusp of developing a true “thinking” machine, but programming a computer to naturally process language, think along abstract lines, and recognize itself proved to be above the technological capabilities of the time. Even after great strides were made in AI during the late twentieth and early twenty-first centuries, true AI remained in the theoretical stage. Although many elements of modern life use AI, it is considered to be narrow or “weak.” Narrow AI is programmed with a limited range of functions, and although it can adapt and learn within its programming, it is still confined by the parameters it is given by humans. Strong AI is self-aware and possesses true human-like intelligence and what could best be described as consciousness.
Overview
Intelligent agents are fundamental elements in modern computing and AI. To be considered an intelligent agent, the agent must first be able to observe and perceive its environment. It must do so autonomously, without any human interaction. It must then process its observations to proactively make decisions. These decisions must result in rational actions to achieve a specific goal. Agents collect input from their environment using sensors such as microphones or cameras. An actuator, such as a display screen or a speaker, delivers the output to the user.
The simplest intelligent agents are given a set of parameters by a human user. These parameters can be a keyword or the date that something was published. The agent then searches the Internet or another database to gather information and provide the answers to the user. An example of this type of intelligent agent is the push notification. If a user opts into such notifications, an intelligent agent will generate an alert or pop-up message when the desired content is available.
Computer scientists generally divide intelligent agents into five types. Simple reflex agents ignore the past and only operate in the present. They work according to the event-condition-action (ECA) rule in which a user initiates an event, and the agent then operates under pre-set conditions to return an action with a preprogrammed outcome. Model-based reflex agents also work along the ECA principle, but their search environment is more comprehensive and includes some of the agent’s historical data.
Goal-based agents use the information stored by a model-based agent and expand upon it by searching for the most desired outcomes. Utility-based agents take the goal-based concept and rate possible outcomes, such as looking at the probabilities of success or the amount of resources needed to accomplish the goal. The agent then takes action to choose a scenario that maximizes the outcome. Learning agents receive and analyze feedback to gradually improve their decision-making process.
Intelligent agents are used in many aspects of the modern world. Intelligent personal assistants such as Alexa or Siri can help users with various daily tasks. These can range from answering simple questions, scheduling appointments and issuing reminders, playing requested music, or making sure lights are turned on or off.
Autonomous robots use intelligent agents to operate independently of humans. A well-known example is the robot vacuum cleaner the Roomba. The Roomba’s goal is to clean a room. To do this, it uses cameras, bump sensors, and dirt sensors to survey its environment. It acts by using its wheels, brushes, and suction power to achieve its goal. Similar types of robot agents include self-driving cars and drone delivery robots.
Fraud detection agents observe bank account and credit card information and flag anything suspicious to alert the user to possible illegal activity. Some traffic lights use intelligent agents to manage traffic flow, changing lights to green when it detects a car and adjusting the duration of light based on the time of day.
Bibliography
Bansall, Sahil. “Agents in Artificial Intelligence.” Geeks for Geeks, 5 June 2023, www.geeksforgeeks.org/agents-artificial-intelligence/. Accessed 22 Aug. 2023.
Hashemi-Pour, Cameron. “Intelligent agent.” TechTarget, 2023, www.techtarget.com/searchenterpriseai/definition/agent-intelligent-agent. Accessed 22 Aug. 2023.
“Intelligent Agent (IA).” A.I. For Anyone, www.aiforanyone.org/glossary/intelligent-agent. Accessed 22 Aug. 2023.
Lee, Rachael. “Hey Siri, Where Are You From? The History Behind Voice Assistants.” Badger Herald, 10 Nov. 2020, badgerherald.com/artsetc/2020/11/10/hey-siri-where-are-you-from-the-history-behind-voice-assistants/. Accessed 22 Aug. 2023.
Oppy, Graham, and David Dowe. “The Turing Test.” Stanford Encyclopedia of Philosophy, 4 Oct. 2021, plato.stanford.edu/entries/turing-test/. Accessed 22 Aug. 2023.
Terra, John. “Exploring Intelligent Agents in Artificial Intelligence.” Simplilearn, 10 Aug. 2023, www.simplilearn.com/what-is-intelligent-agent-in-ai-types-function-article. Accessed 22 Aug. 2023.
“What is Artificial Intelligence (AI)?” IBM, www.ibm.com/topics/artificial-intelligence. Accessed 22 Aug. 2023.