“We can make use of AI even when developing AI systems”

The Bizden Haberler Podcast series moderated by journalist Cüneyt Toros this time welcomes Dr Metin Sezgin, a professor of computer engineering at Koç University’s engineering faculty. Dr Sezgin, who also serves on the Koç University İş Bankası Artificial Intelligence Application & Research Center (KUIS) Committee, says that artificial intelligence technologies are likely to have a far greater impact on jobs that are repetitive, don’t require a lot of thought, and are more easily automated.

The Bizden Haberler Podcast Series’ most recent guest was the distinguished computer scientist Professor Metin Sezgin. Dr Sezgin has been pursuing an academic career at Koç University’s computer engineering department for 15 years but his research on artificial intelligence technologies stretches back over 25 years. Professor Sezgin graduated with honors from Syracuse University in 1999, then went on to earn his master’s degree in 2001 and his doctorate in 2006, both from the Massachusetts Institute of Technology (MIT). After joining the University of Cambridge Computer Laboratory as a postdoctoral research associate in 2006, he joined Koç University in 2009. Dr Sezgin currently holds a position in the Department of Computer Engineering and serves as a member of the Koç University İş Bankası Artificial Intelligence Application and Research Center Committee. In addition to his other roles, he is also the vice president of the Turkish Artificial Intelligence Technology Association, of which he was one of the founders.

Could you tell us about the KUIS Artificial Intelligence Application and Research Center and your team there?

I’ve been with the Computer Engineering Department at Koç University for 15 years now. I’ve been working on artificial intelligence since the day I arrived. In those early years, there were just a handful of us focused on artificial intelligence. As time went on, a team of people investigating AI began to form. About six or seven years ago the notion of creating an AI laboratory also took shape, the idea being that this would make Koç University a center of attraction in this field and our team would grow even more as a result. With the invaluable support of KU’s provost and İş Bankası, we established the Koç University İş Bankası Artificial Intelligence Application and Research Center.

We currently have a large number of faculty members working on various projects at our center such as language models, human-computer interaction, AI systems that combine text with visuals or other communication elements, robotic systems...

It seems like AI is impacting virtually every industry these days. Research suggests that the financial services industry in particular will experience significant changes. What’s your take on the future that lies ahead?

In its present state, AI relies primarily on data-driven learning. AI will therefore be most useful in areas where there is a lot of data and we can draw conclusions from that data to make people’s lives easier. In any field where we can leverage data for predictions, particularly those with challenges we haven’t yet overcome, AI has the potential to make a substantial impact. Finance is perhaps the most salient example, but healthcare and education are also seeing a lot of exciting developments in AI. I’m also the vice president of the Turkish Artificial Intelligence Technology Association, of which I was one of the founders. One of our most active working groups at the association is finance; another is healthcare.

Accessing data in these fields can be a bit more challenging compared to others. It’s particularly difficult in healthcare because the information is so heterogeneous. AI will undoubtedly make significant contributions to education as well.

There’s also the industrial and manufacturing side of generative AI to consider. Which sectors are likely to undergo the most profound transformation, and what do you believe will emerge in their place?

Jobs that are largely repetitive and routine in nature are the ones that will disappear first. It used to be that developing an AI system involved significant human effort but nowadays AI systems are being used to build new AI systems. In other words we can make use of AI even when developing AI systems. As for generative AI, large language models have yielded impressive outcomes, especially when it comes to generating written text. It’s important to note however that these “creative” generative AIs aren’t really creating something new. They’re learning from and mimicking existing works, effectively appropriating the labor of countless individuals and potentially displacing them from their jobs. That’s another reality we have to face.

Considering all the privacy, security, and potential psychological impact issues that may be involved, how should AI communicate with children?

I believe children’s interactions with technology need to be regulated and kept under control. Children’s interaction with artificial intelligence, in particular, raises concerns. In one of my research areas, the development of emotionally intelligent robots, we see that these technologies can almost be perceived as human because they can reflect human-like emotions, “read” the emotions of the person they’re interacting with, speak like a human, understand human speech, and provide reasonable responses.

This allows them to both influence and be influenced. It therefore remains an open question how people of all ages–whether children or adults–will interact and coexist with AI systems, how they will make decisions using these systems, or how financial and commercial institutions might be manipulated by the companies that create this software and the robots that the software powers. Just as we once used to talk about concepts like media literacy in the context of social media and media consumption, we now need to also keep such issues as these in mind when engaging with interactive AI systems.