Surely humans do not act as machines - for they are no machines. This applies for daily human behavior as well as human working environments. Human behavior is not always fully linear, not completely predictable at every stage of the working process and sometimes apparently inefficient. Real people make exceptions where machines strictly follow the rule. In former times you would have regarded these facts as a sign of human intelligence. These obvious platitudes can become quite doubtful when human workers are more and more controlled by electronic devices and software programs. So the still imperfect but at the core human modern working environment more and more collides with the demand for perfectly efficient working processes. These completely controlled processes are regarded as a guarantor for economic success at a progressive rate. With regard to algorithms controlling more and more working environments some questions should be allowed about the ethical implications of this development. A notorious example is Amazon and its sophisticated computer system supervising and leading all factors in the delivery process including the human work force. Is it right when machines set the metrics of human work? Does this development truly guarantee more economic success than original human work and motivation? The working environment is not the only example of machines grasping for control - just have a look on automatized processes taking over control in your car. We the robot in your car make better decisions and will traffic become more predictable and safer? We doubt this.
What we do here is to strike a blow for more humanity among algorithms.
Efficiency versus humanity?
The Amazon system where workers are controlled and directed by a supervising computer system is by far not the only example of algorithms in control. But the Amazon process is very significant when it comes to open up the implications of such an approach especially in a working environment. One clear finding is that it is "ruthless approach" (1) to treat humans like robots by discarding obvious ineffective workers without hesitation. It leaves a work force which is not led by inner motivation but by a "machine god". Employers who rely on such methods will not be able to build up a stable work force which clings to the firm for a long time. The costs of replacing workers very often and to break the newcomers in have to be put on the bill too. Also treating workers as "ants" will only be successful as long enough free workers play in the market. If businesses get into an economic situation where there is more work than people a more human treatment of the work force will be the key to keep the workforce. For some countries in Europe this situation is not as farfetched as it sounds because of the gap in their demographic development. So we regard a machine ruled workers treatment as economically shortsighted and ethically questionable.
Efficiency Gaps in Machine Routines
Also it can be doubted that machine controlled processes always persuade with more efficiency. Amazon offers just another example of machine routines in connection with human tasks which leave some doubts about such processes. We talk about the Amazon Mechanical Turk which offers a marketplace for work. A study of Panagiotis G. Ipeirotis and others, Leonard Stern School of Business (2), finds some remarkable flaws in the machine control especially when it comes to assess the individual worker's performance. At this point the efficiency routine becomes a disadvantage as the typical machine routine is unable to have a closer look on the individual case. The authors of the study offer a new code to solve some of the problems but it shines through that just changing some algorithms is not the solution to the core problem: Machines make no exceptions and can just perform the way their programs run.
Chances of Human-Computer Interaction
No doubt, when machines serve humans the human-computer interaction offers a lot of chances as well. As a study of José J. Cañas and others points out, ergonomics in the working environment are supported by intelligent machine solutions when the machine remains the servant and the human is the leader. These means that an intelligent interaction between machines and humans still somehow leaves the human in the driver's seat. Especially Software as a Service (SaaS) solutions which come from outside into the enterprise should be designed to this basic principle. These software solutions from outside becoming the ruler over human people is always at risk not to support the typical chemistry which makes a business successful because as a machine it lacks human intuition.
Why Human Intelligence is different
Other ethical dilemmas concern intelligent machines which have to make human decisions. It is understood that an intelligent car replacing the human driver will have to be programmed to kill. (4) Even cool algorithms get into moral difficulties when they face a not avoidable accident. What if the machine can just choose between two possibilities: Just kill two people or 20 people, no other solution given? In a minor form this is the often awful decision a human driver has to make in a second when forced between killing a dog running in his car or risking the life of the driver behind by stopping abruptly? Yes, algorithms have to be very sophisticated to get into such sensitive and complex decision-making spheres. Scientists and engineers work on such solutions which make clear that the human element is not dispensable. It is a part of intelligence because intelligence is not just algorithm.
Having this in mind there is no limit to a very successful human-machine-interaction which makes human life more comfortable, safer and sometimes even more human. Humanity remains the metrics of machine ruling and the Software ideally even understands human behavior. At least a bit.
References:
(1) http://edgardaily.com/en/life/2015/inside-amazon-how-computers-rule-over-human-employees-28131
(2) http://www.ipeirotis.com/publications - A Framework of Quality Assurance in Crowdsourcing
(3) Human Factors and Ergonomics, José J. Cañas1, Boris B. Velichkovsky2 and Boris M. Velichkovsky 3, University of Granada, Spain, Kurchatov Research Institute, Moscow, Russian Federation, Dresden University of Technology, Germany
(4) http://www.technologyreview.com/view/542626/why-self-driving-cars-must-be-programmed-to-kill/
What we do here is to strike a blow for more humanity among algorithms.
Efficiency versus humanity?
The Amazon system where workers are controlled and directed by a supervising computer system is by far not the only example of algorithms in control. But the Amazon process is very significant when it comes to open up the implications of such an approach especially in a working environment. One clear finding is that it is "ruthless approach" (1) to treat humans like robots by discarding obvious ineffective workers without hesitation. It leaves a work force which is not led by inner motivation but by a "machine god". Employers who rely on such methods will not be able to build up a stable work force which clings to the firm for a long time. The costs of replacing workers very often and to break the newcomers in have to be put on the bill too. Also treating workers as "ants" will only be successful as long enough free workers play in the market. If businesses get into an economic situation where there is more work than people a more human treatment of the work force will be the key to keep the workforce. For some countries in Europe this situation is not as farfetched as it sounds because of the gap in their demographic development. So we regard a machine ruled workers treatment as economically shortsighted and ethically questionable.
Efficiency Gaps in Machine Routines
Also it can be doubted that machine controlled processes always persuade with more efficiency. Amazon offers just another example of machine routines in connection with human tasks which leave some doubts about such processes. We talk about the Amazon Mechanical Turk which offers a marketplace for work. A study of Panagiotis G. Ipeirotis and others, Leonard Stern School of Business (2), finds some remarkable flaws in the machine control especially when it comes to assess the individual worker's performance. At this point the efficiency routine becomes a disadvantage as the typical machine routine is unable to have a closer look on the individual case. The authors of the study offer a new code to solve some of the problems but it shines through that just changing some algorithms is not the solution to the core problem: Machines make no exceptions and can just perform the way their programs run.
Chances of Human-Computer Interaction
No doubt, when machines serve humans the human-computer interaction offers a lot of chances as well. As a study of José J. Cañas and others points out, ergonomics in the working environment are supported by intelligent machine solutions when the machine remains the servant and the human is the leader. These means that an intelligent interaction between machines and humans still somehow leaves the human in the driver's seat. Especially Software as a Service (SaaS) solutions which come from outside into the enterprise should be designed to this basic principle. These software solutions from outside becoming the ruler over human people is always at risk not to support the typical chemistry which makes a business successful because as a machine it lacks human intuition.
Why Human Intelligence is different
Other ethical dilemmas concern intelligent machines which have to make human decisions. It is understood that an intelligent car replacing the human driver will have to be programmed to kill. (4) Even cool algorithms get into moral difficulties when they face a not avoidable accident. What if the machine can just choose between two possibilities: Just kill two people or 20 people, no other solution given? In a minor form this is the often awful decision a human driver has to make in a second when forced between killing a dog running in his car or risking the life of the driver behind by stopping abruptly? Yes, algorithms have to be very sophisticated to get into such sensitive and complex decision-making spheres. Scientists and engineers work on such solutions which make clear that the human element is not dispensable. It is a part of intelligence because intelligence is not just algorithm.
Having this in mind there is no limit to a very successful human-machine-interaction which makes human life more comfortable, safer and sometimes even more human. Humanity remains the metrics of machine ruling and the Software ideally even understands human behavior. At least a bit.
References:
(1) http://edgardaily.com/en/life/2015/inside-amazon-how-computers-rule-over-human-employees-28131
(2) http://www.ipeirotis.com/publications - A Framework of Quality Assurance in Crowdsourcing
(3) Human Factors and Ergonomics, José J. Cañas1, Boris B. Velichkovsky2 and Boris M. Velichkovsky 3, University of Granada, Spain, Kurchatov Research Institute, Moscow, Russian Federation, Dresden University of Technology, Germany
(4) http://www.technologyreview.com/view/542626/why-self-driving-cars-must-be-programmed-to-kill/
RSS-Feed