Artificial intelligence
Author : DAOUDI Samir | Context : MSc Software Engineering – Computer Structure |
A real challenge for humans: the artificial intelligence. Is it possible for a
machine to think? This question was the source of a lot of debates. The community of scientists is divided into two major opinions. Some scientists think that it is or it would be possible in the future for a machine to think and act as human mind or better. Others are against this opinion and are convinced that machine will simply act as it has been programmed.
Defining what AI is? “Intelligent behavior involves perception, reasoning, learning, communicating and acting in a complex environment” [1].
If we try to seek more details and try to define what are minds or what is thinking? This kind of questions have interested for long time philosophers [2].
As any innovation has its Pros & Cons this also applies to AI. Responding to the question of dangers involved, I would respond it depends on the domain of application. What can be dangerous is the fact of giving the machine the ability to act and make critical decision.
The basic concept of intelligence is implemented in a unit called agent that can receive inputs from its environment and interact with it. We can definitely say that an agent is intelligent if:
– What It does is appropriate for the circumstances.
– It is easy to change its goal by changing the environment.
– It can increase its knowledge by learning from previous situations.[3]
My personal point of view is even if we attend a high level of artificial intelligence and we will be able to program very smart machines, those should be kept for helping us the humans and not doing the role of humans.
Robots or machine that are used for the tracking system, generating alerts, sending notifications or having control over a limited set of non critical tasks can provide a great help and support for us and can make our life easier. For this situation I would give a concrete example from my experience. In our company there is a critical job that is the Pipe Controller. It consists in controlling different values related to the pressures, temperature, quantity of gas…etc within the pipe. Depending of certain values different actions can be taken as activating process, notifying specific intervention team, start an urgent shut down of turbines …etc. Three teams of 6 persons, each team covers 8h/day with non-stop tracking and visualizing graphs and reports. Since last year a new system has been deployed and a great part of these tasks got automatized and the work of humans has been significantly facilitated thanks to the application. The solution provided a lot of ‘intelligence’ by letting it learn and predict future situation and solutions. The teams took a great decision they disabled those features to avoid any inconsistent decision that would be fatal.
This is a situation where humans use the intelligence of a machine with restricted power. And I think this attitude should be adopted with every ‘intelligent’ machine. Yes it may present a danger especially in the case where it is allowed to learn, enlarge its knowledge database and take decision.
References
[1] Nils J. Nilsson, ‘Artificial Intelligence: a new synthesis’, 1998. ISBN: 1-55860-555-5.
[2] John Haugeland ,‘Artificial intelligence: the very idea’, 1989. ISBN:0-262-58095-0.
[3] David L. Poole, Alan K. Mackworth , ‘Artificial Intelligence: Foundations of Computational Agents’