Founding Director Azer Bestavros Writes on His Vision for Ethical Artificial Intelligence

Founding Director of the Hariri Institute for Computing Professor Azer Bestavros wrote an article featured in BU Today called “Artificial Intelligence Has a Powerful Brain, but It Still Needs a Heart.” In the world of growing technology, Professor Bestavros argues that we need to make sure artificial intelligence being adapted is ethical and beneficial for all parties involved and without discrimination.

However, the enthusiasm to apply the up and coming technology is greater than the efforts to cultivate fairness. For this reason, higher educational institutions are responsible for taking the lead in creating “an AI-powered world that is both ethical and fair.” Ethics officers and review boards are being established, but those alone are not enough to ensure ethical practices. Artificial Intelligence can only base its results on the data it’s given. This creates a bias because the data handed to machines are the result of a series of human decisions. As Professor Bestavros says, “Algorithms do not self-correct. They self-reinforce.”

Many universities like Boston University, Harvard, Columbia, and Berkeley have jointly taught a Law for Algorithms course, and other universities, like Cornell and MIT, offer courses on the ethical design of intelligence systems. Universities need the government’s help to understand learning machines and how they come to their conclusions, so they can explain their results to the communities the information affects. Laws and regulations support the universities’ research and require industries to comply, so it’s a win-win partnership. Professor Bestavros concludes that people in government, academia, and industry must band together to develop and apply ethical artificial intelligence, but that higher education must lead to move forward.

Azer Bestavros is the Founding Director of the Hariri Institute for Computing at Boston University. He is also a Professor in the Department of Computer Science in the College of Arts and Sciences and holds affiliated appointments in Electrical and Computer Engineering and in Systems Engineering. His current research is focused on the design, implementation, and deployment of scalable software platforms for privacy-preserving big-data analytics, and the use of edge clouds for control of cyber-physical systems.