Faculty of Engineering and Natural Sciences Dean Prof. Dr. Arıca Evaluates Artificial Intelligence Technology

Evaluating developments in artificial intelligence technology, Bahçeşehir University Faculty of Engineering and Natural Sciences Dean Prof. Dr. Nafiz Arıca said: "I believe it may reach the potential to threaten humanity in the 2050s."

A TÜBİTAK-SUPPORTED PROJECT

Prof. Dr. Nafiz Arıca talked about his TÜBİTAK-supported artificial intelligence project.
Describing the project "Expression Analysis Based on High-Level Facial Expression Features and Deep Learning Architectures in Video Footage Recorded in Unconstrained Environments", Arıca said: "In this project, we try to analyze the expressions on people's faces by tracking their faces in video footage containing more than one person. The project has reached a certain point."

"POTENTIAL CRIMINALS WILL BE IDENTIFIED IN ADVANCE"

Stating that the project can also be used for security purposes, Prof. Dr. Arıca said: "Suspicious movements will differ from ordinary movements. It is not at all easy for an operator watching video footage in a shopping mall to watch hundreds of images one by one, foresee criminals and take precautions in advance. At the very least, operator errors come into play. In this sense, such analysis algorithms and image algorithms will make it possible to identify potential criminals in advance and at least issue a warning."

"ONE OF THE MOST IMPORTANT ELEMENTS OF SECURITY IS TAKING PRECAUTIONS IN ADVANCE"

Saying "In the not-too-distant future, systems will perhaps be developed in which glasses given to law enforcement officers will issue a warning about wanted people in their field of view," Arıca noted that there are already some examples of this in the world.

Nafiz Arıca said: "In this way, not only wanted people but also potential criminals and suspicious movements will be detected automatically and precautions will be taken in advance. One of the most important elements of security is taking precautions in advance and detecting in advance."

"ARTIFICIAL INTELLIGENCE MAY REACH THE POTENTIAL TO THREATEN HUMANITY IN THE 2050s"

Drawing attention to the danger in artificial intelligence technology, Prof. Dr. Nafiz Arıca said: "There are many applications with artificial intelligence features. In the past five years they have reached a very serious scale and significantly affected our lives. In the long term, I believe they will have major effects. I believe it may reach the potential to threaten humanity in the 2050s. I believe it will continue with this momentum over the next 5-10 years and then reach a plateau, and that with developments in the 2050s, systems will emerge that can threaten humanity: systems that can think for themselves, interpret and reason, but that can also become a threat by learning things humanity does not want."

"IT MAY THREATEN HUMANITY IN THE MEDIUM TERM"

Arıca stated that machines may become a new species in the long term. Nafiz Arıca said: "The reason artificial intelligence has advanced so much in the past 5-10 years is the development of the 'deep learning' approach. Instead of the features we used to give by hand in artificial intelligence algorithms such as image and sound classification, I am now talking about 'deep learning' algorithms learning automatically and without supervision from large amounts of data. As a result, in the long term these systems may also learn things we do not want. By learning things we do not want, there may be systems that behave and act in ways we do not want. These are predictions, but I believe that even in the medium term it will be a technology that may threaten humanity, with both positive and negative aspects."

"IN MY OPINION, IMPOSING SANCTIONS ON ARTIFICIAL INTELLIGENCE SYSTEMS IS OUT OF THE QUESTION"

Also speaking about a driverless car hitting a pedestrian in America, Arıca said: "A driverless vehicle killed a cyclist. There are many debates like this. In my opinion, it is not possible to hold an artificial intelligence system morally responsible. In these debates, an entity's moral responsibility generally comes from having two attributes. One is intellect, the ability to reason. The other is sentience, that is, emotional attributes: the potential to feel pain. The artificial intelligence system, the mind of that driverless vehicle, exists in a sense, but it has no capacity to suffer. That is why, in my opinion, imposing sanctions on such artificial intelligence systems is out of the question. Let us set that aside for now. This debate will take place much more on legal grounds in the coming years."

"HOW KNOWLEDGE IS USED IS WHAT MATTERS"

Nafiz Arıca said: "Access to information is now very easy. At Bahçeşehir University, we strive to teach our students how to use knowledge. Because access to information is so easy, how knowledge is used is what matters. This involves communication, design and ethics. At Bahçeşehir University Faculty of Engineering and Natural Sciences, we strive to educate our students not merely as engineers but as people with strong communication skills and innovative and creative thinking abilities."