Intelligent agents and artificial intelligence

Human-computer interaction or Human-computer interaction is widely studied in many fields of research. Possible applications range from logical information access systems, e-commerce services, e-learning tutoring for language teaching, through the animation of avatars in the virtual environment and computer games.

But is it possible to build completely autonomous agents? An intelligent robot must not only imitate human actions, but must also behave like a real person. Eye movements, facial expressions, appropriate gestures - it is very difficult to catch all these nuances in an artificial system. (1)

One of the branches that is beginning to gain more popularity today globally, in Artificial Intelligence, is that of Intelligent Agents. An autonomous agent is considered one that is a system located in and as part of an environment, sensing that environment and acting on it, over time, trying to achieve its own objectives in a way that affects what it feels in the future.

An agent is anything that can be considered to perceive its environment through sensors and to respond or act in that environment through effectors or actuators. Human agents have eyes, ears and other organs that use sensors, as well as hands, legs, mouth and other parts of the body that use effectors. In the case of robotic agents, the sensors are replaced by infrared cameras and the effectors are replaced by motors. For a software agent, the perceptions and actions will be the encoded bit strings. (2)

Agents can be classified according to the environment they are in, e.g. Intelligent Agents, Biological Agents, Software Agents, Hardware Agents, Interface Agents, Mobile Agents, Reactive Agents, Information Agents, Agents based on Distributed Artificial Intelligence DAI...

An intelligent agent is more than just an information search system. The agent can operate without user intervention even if the user is offline. Intelligent agents incorporate functions from artificial intelligence.

An agent must strive to “do the right thing,” based on what it can perceive and the actions it can take. The right action is the one that will make the agent more successful.

The performance measure is the objective criterion for the success of an agent's behavior.

For example, the performance measure of a cleaning agent could be the amount of dirt cleaned, the amount of time consumed, the amount of electricity consumed, the amount of noise generated, etc.

For each possible perceived sequence, a rational agent must select an action that is expected to maximize its performance measure, given the evidence provided by the perceived sequence and any embodied knowledge the agent has.

Rationality is different from omniscience (all knowledge with infinite knowledge). Agents can take actions to modify future perceptions to obtain useful information (information gathering, exploration)

An agent is autonomous if its behavior is determined by its own experience (with the ability to learn and adapt)

PEAS: Performance measure, Environment, Actuators, Sensors.

To set the PEAS, the configuration for the intelligent agent design must first be specified.

Consider, for example, the task of designing an automated taxi driver and establishing the Performance Measure Environment Actuators Sensors:

Configuration must first be specified for the intelligent agent design For example, the task of designing an automated taxi driver would involve the following PEAS (SELF DRIVING CAR):

Performance measure: safe, fast, legal, comfortable ride, maximize profits

Environment: Roads, other traffic, pedestrians, customers Actuators: Steering wheel, accelerator, brake, signal, horn

Sensors: Cameras, sonar, speedometer, GPS, odometer, engine sensors, keyboard, etc...

But what guarantees that robots in this 4th industrial revolution driven by robotics and artificial intelligence do "the right thing"?

The three laws of robotics (3) are a set of rules written by Isaac Asimov, which most of the robots in his novels and stories are designed to follow. In that universe, the laws are "mathematical formulations printed in the positronic brain pathways" of the robots (lines of code from the program that regulates compliance with the laws stored in the robot's main memory). Appearing for the first time in the story "Runaround" (1942), they establish the following:

1. A robot shall not harm a human being or, through inaction, allow a human being to come to harm.
2. A robot must do or carry out the orders given by human beings, except if these orders conflict with the 1st Law. A robot must protect its own existence to the extent that this protection does not conflict with the 1st or 2nd Law.
3. A robot must protect its own existence to the extent that this protection does not conflict with the 1st or 2nd Law. References: 1. Fernando Jiménez Motte , “Multilanguage Platforms, Intelligent Agents, Artificial Intelligence AI and Artificial Language” 2. Christian Crovetto Artificial Intelligence and Robotics. 3. The 3 Laws of Robotics, Wikipedia

Fernando Jiménez Motte Ph.D (c) EE, MSEE, BSEE

CEO of NEUROMORPHIC TECHNOLOGIES NT Robotics, Control Systems, Artificial Intelligence AI

https://fernandojimenezmotte.com/

https://neuromorphic-technologies.com/

Follow me on Twitter @stockfjm