Download the Current Issue
Dear ComputingEdge reader:
Values in AI Software: Artificial intelligence (AI) attempts to imitate human behavior and reasoning, but what about ethics and morality? As AI becomes more integrated into our lives, there is a growing consensus about the importance of instilling human values such as fairness and respect in AI software. This issue of ComputingEdge focuses on strategies for creating ethical AI software.
Avoiding Simpson’s Paradox
Heard of Simpson’s paradox, gentle colleagues? I know some of you have. It’s from data science, and it reminds us that our systems can be inadvertently biased, favoring one group over another even when we are trying our best to be just and fair. In short, Simpson’s paradox says that it is possible—actually quite easy—to make individual engineering decisions that are completely free from bias and yet end up with a system that is as skewed as a drunken bigot on a tear. If we are to build honest, fair, and unbiased systems, dear colleagues, we must be vigilant. Three articles in this month’s ComputingEdge remind us just how vigilant we must be.
Seventeen authors take up this question directly in “Think Your Artificial Intelligence Software Is Fair? Think Again.” They look at the basic, common engineering tasks that can subtly twist the work that we are doing. “There is a lot of work left to do to achieve unbiased AI,” they conclude. “Fairness is a multifaceted, context-dependent social construct that defies simple definition.” It depends on your technology, your algorithms, your data, and ultimately your view of your neighbor.
Next, we have “Artificial Intelligence for Law Enforcement: Challenges and Opportunities,” by Stephan Raaijmakers. Raaijmakers notes that we have demonstrated many tools that might be of use to law enforcement. Artificial intelligence “offers great opportunities for accurate data analysis and interpretation.” Yet he notes that “a number of important problems will have to be solved for successfully integrating” these technologies into law-enforcement workflows.
Finally, we have a reflective piece about artificial intelligence: “From Artificial Intelligence to Artificial Wisdom: What Socrates Teaches Us,” by Tae Wan Kim and Santiago Mejia. It’s a provocative and thoughtful piece that reminds us that we must reflect on what we are, especially when our systems cannot. “The absence of sustained reflection and critical perspective is seriously hindering any healthy democratic deliberation.” You can start reflecting by reading this month’s ComputingEdge.
—David Alan Grier for ComputingEdge