The technological advances of the last few decades have revolutionized our way of life, but it has become clear that many innovations have sacrificed ethics, and what might be beneficial for some comes at a cost for many others.
A new study, conducted by North Carolina State University, tackles the question of how important codes of ethics are to software developers. Their conclusion: ethics doesn’t affect the decisions made by these people. The work came after the Association for Computing Machinery (ACM), the world’s largest computing society, updated their code of ethics last July.
"We applauded the decision to update the ACM code of ethics, but wanted to know whether it would actually make a difference," co-author Professor Emerson Murphy-Hill said in a statement. "This issue is timely, given the tech-related ethics scandals in the news in recent years, such as when Volkswagen manipulated its technology that monitored vehicle emissions. And developers will continue to face work-related challenges that touch on ethical issues, such as the appropriate use of artificial intelligence."
The study involved 168 people, 105 of whom were US software developers with five or more years experience in the field and 63 of whom were software engineering graduate students at university. Half of the participants were given the ACM code of ethics, while the others were just told that codes of ethics were an important part of the test. They were all given 11 written scenarios involving ethical challenges and were asked how they would respond.
"There was no significant difference in the results – having people review the code of ethics beforehand did not appear to influence their responses," Murphy-Hill added. "While we believe maintaining an up-to-date, robust code of ethics is an admirable thing for ACM to do, we were unable to find any effect of the code of ethics on developer decision making. The question now becomes: What can the computing profession do to promote ethical behavior?"
Ethics in software development, and in technology in general, has rightly come under intense scrutiny in the last few years. From privacy scandals across many social networks to algorithms that have the same biases as humans, there is a desperate need for improvement.
One limitation of the study was that the questionnaire only presented two possible actions for each scenario (as well as an “unsure” option). "Decisions in these situations rarely have only two available actions, therefore, some participants might be unsure what they would do because they would really do neither of the options presented," wrote the authors.
Nonetheless, hopefully when this work is presented next month at the ACM Symposium on the Foundations of Software Engineering, it can start some wide-reaching and meaningful changes.