Artificial intelligence and its building block

Artificial intelligence, one of the most commonly used term these days. Before venturing into AI, let us understand the evolution of humankind.

What makes Homo sapiens so unique among all the animals? It’s the ability to think, we have evolved from the Homo habilis stage considered to be the first humans emerged to be known.

As time progressed, we evolved ourselves by using tools such as stones, woods, and bones. Our intelligence enabled us towards many path-breaking innovations.

Artificial intelligence evolution

In today’s world, we come across the most commonly used term Artificial intelligence or AI. AI attempts not just to understand but also to build intelligent entities. Before understanding the intelligent agents, let us know the basic definition of artificial intelligence.

Artificial intelligence defined into two different perspectives based on thought process & reasoning, and behavior.

Categorized into four different approaches, as shown in the table below.

Artificial intelligence definition

Acting humanly, also known as the Turing test approach. A system passes the test if the interrogator cannot distinguish whether the answer is from a human or a computer, then the system possesses artificial intelligence.

Thinking humanly, known as the cognitive model approach when a system or a program thinks like a human. But we need to understand how humans think. To be determined through introspection, psychological experiments, and brain mapping when the program’s behavior matches the human behavior in solving the same problems.

Thinking-rationally, known as the law of thought approach when a system that could solve any solvable problem describe in logical notation that could involve the use of computational models.

Acting rationally: The rational agent approach, a system which operates autonomously, perceive their environment and adapt to achieve the best outcome.

What is an intelligent agent?

Agent that can distinguish or recognize its environment through sensors and acting upon the environment through actuators. The behavior of the agent described as an agent function, which would match the percept sequence to an action. The agent function for an AI system implemented through an agent program, running within the physical system.

Artificial intelligence environment

 

Let us consider a simple example of an autonomous object mover in the manufacturing industry. The environment involves two locations: – location A and location B.

The autonomous object mover performs the following agent function, if the current location has the object, then pick up the commodity, otherwise move to other location.

Artificial intelligence object mover

Percept Sequence Action
[A, empty] Right
[A, object] [Pick the object]
[B, empty] Left
[B, object] Pick the object

Simple agent function for the object mover, interacting with the environment disclosed above. A rational agent is one that does the right thing when a sequence of action is desirable. The sequence of actions causes the environment to go through an array of states. The design performance must measure according to how an agent must behave in the surrounding. When we relate the object mover in manufacturing industries, performance is measured based on the movement of the object.

 

Geography of the environment known as priori, the distribution of the objects and location doesn’t come under priori. The available action for the rational agent is left, right, and move the object. The agent correctly perceives its location and identifies whether that location contains a commodity. Under these conditions, we consider the agent is rational. But the same agent would be irrational under different circumstances.

For example, the agent needs to explore the environment when the geography changes. The agent must adjust rather than stick to a sequence of defined events.

Ideal rational agent in reality

In case an agent knows the actual outcome of its action and acts accordingly. Then such kind of agent is considered to be an omniscient agent. But such kind of agent is impossible in reality. In case an agent knows the actual outcome of its action and acts accordingly. Then such kind of agent is considered to be an omniscient agent. Such kind of agent may be rare in reality.

An ideal rational agent must not only gather information but also needs to learn as much as possible from what it perceives. If a rational agent has to cross the road, the agent must look around before crossing. Because looking around maximizes the expected performance. Action by agent during the event sequence to modify future percepts known as information gathering.

A rational agent must be autonomous and should learn how to compensate for partial or incorrect prior knowledge. It would be reasonable to provide an agent with some initial knowledge and ability to learn.

For any agent, we must specify performance measure, the environment, agent’s sensors, and agent’s actuators. We group this into function blocks required for developing an intelligence agent.

For an object mover in the manufacturing industry, the table below depicts each element.

Fetch robotics are pioneers in the field of freight robotics solutions developed for the warehouse environment. Robots designed to work alongside people and other heavy equipment within the busy warehouse environments.

Artificial intelligence Freight 500

Image Credit: Fetch Robotics

What is agent program?

Action performed by the agent based on the sequence of percepts describes its behavior.

Based on agent’s behavior, the agent program would implement an agent function that helps to map from percepts to actions. This program will run on a computing device integrated with sensors and actuators, defined as system architecture.

The program recommends the action to the computing device such as FPGA, PC, or microprocessor. The actuators and sensors perform the desired action based on the sequence of percepts. The system architecture makes the percept from the sensors available to the program. Runs the program and feeds the programs action choice to the actuators.

 

Disclaimer: The opinions expressed within this article are the personal opinions of the author. Circuit Generator is not responsible for the accuracy, completeness, or validity of any information on this article. Circuit generator does not assume any responsibility or liability for the same.

Leave a Comment