There is no widely agreed definition of AI, but for practical purposes we can say that it is concerned with the design and creation of rational agents. An agent is simply something that does things. The term rational is used to indicate we are not trying to emulate human beings here – where the decision making process can be anything but rational. Of course the term ‘rational’ is predicated on a well understood objective that we want our agent to achieve. A rational agent should do the best it can to make sure the objective is met or exceeded.
An agent has roughly three parts – its sensors, an algorithm or program, and actuators so it can do something. An agent that can recognize faces for example, needs a camera, image processing software, and an output device so it can confirm or otherwise, someone’s face. from Google’s driverless car, through to a robot spraying cars with paint on an assembly line, these three components are needed to define an agent.
Agents operate in various environments. Sometimes these environments are very well defined (the paint spraying robots), and sometimes very poorly defined – the roads that a driverless car has to navigate for example. In the latter case there are many unforeseen things that can happen – a car jumping red lights, a pedestrian stepping out into the road without looking – and so on.
The first robots to appear are those which operate in very well defined environments – manufacturing for example. It is no mistake that the Chinese company that makes the iPhone will be replacing 60,000 workers (yes 60,000) with robots, and Adidas is bringing manufacturing back to Germany and factories employing robots.
AI is a very broad topic, but the three central components of sensors (input), program (processing) and actuators (output), pretty well define an intelligent agent. As time progresses we will come to depend more on agents that learn, and there are already many instances where agents learn at a faster rate than humans and become more accurate decision makers. The rise of AI will not just affect blue collar workers, AI is already being used in medicine and law practice – it will eventually become ubiquitous.