site.btaProf. Robert Tarjan Expects Great Research, Well-Prepared Students from Bulgaria's INSAIT

In an interview for BTA's Antoaneta Markova in Heidelberg, Prof. Robert Tarjan said what he expects from the Institute for Computer Science, Artificial Intelligence and Technology (INSAIT) in Sofia is great research and exceptionally well-prepared students. Prof. Tarjan, who is a laureate of two of the most prestigious prizes in mathematics and computer science - Turing and Nevanlinna, was among the participants of the 11th Heidelberg Laureate Forum from September 22 to 27.

Prof. Tarjan teaches at Princeton University and was in Sofia several months ago for the launch of a scientific and research department at INSAIT in the field of algorithms. He has written some of the key and most widespread algorithms in mathematics and computer science.

"I believe that the goal is to set a strong educational programme in high schools and universities, and students to stay in Bulgaria," he told BTA. "They have high aspirations and chances of success. Some of my colleagues are in Sofia. I am actively working with them and I expect to return to Sofia," he added.

"Think, develop your skepticism, read carefully, ask questions about everything," Prof. Tarjan replied when asked what are the most important skills young people need in the 21st century. The mathematician advises young people to discover what they love to do, "because if you want to succeed at something, you have to be willing to work really hard and fail a lot. Don't be afraid of failure." 

Prof. Tarjan often reminds his students that in mathematics, the answers are yes or no, but one never knows in advance what is the right answer, so one must try and find the solution. "Because if you're wrong and you thought the answer was 'no,' you'll learn something that will help you get to the right answer, 'yes,' and vice versa. I think active learning is very important," he explained. "I try to give students challenge problems, not just to do calculations," Prof. Tarjan told BTA. He did not teach for a long time while working at Bell Labs, and after the first year he felt he seemed to lose his motivation for research. Working with active young minds is very motivating for research, he underscored.

The author of numerous algorithms, which are also applied in computer science, is adamant that it is not algorithms that should verify and authenticate the work of artificial intelligence (AI). "I think that's a job for humans. They should verify the results, humans should be involved in the process. I think it's absolutely impossible to certify AI performance without human involvement. I'm not an AI expert, I'm an algorithm specialist and that's my personal opinion. I'm a bit skeptical about the possibility of algorithms verifying algorithms," he said. The mathematician does not rule out the possibility of creating an algorithm that verifies the results of a computational algorithm. "But AI is a system that is not designed to do computations, but to perform activities that are intrinsic to humans. In it, the answers are not always yes or no, black or white. AI is about the quality of decisions. Is it efficient for human situations? Is it useful for human beings? I think people have to make those decisions," Prof. Tarjan said, stressing that "this is the personal opinion of someone who is not in the AI field." 

Early in his career in mathematics and computer science, he did not give much thought to such ethical issues. 

"Algorithms were a good thing, weren't they? But now, in certain contexts, they're becoming a dirty word because there are algorithms making decisions about human life," Prof. Tarjan said.

"AI algorithms decide whether you get a loan or not, whether you get a bank account or not, determine the length of your sentence when you're convicted. I think it's very important to have humans involved in this process. If decisions are being made that affect people's lives, we should trust AI, but we should check," he said. He noted that there has been a lot of work recently on AI systems explaining how they arrive at a decision or proposal. We get a conclusion, a recommendation from the system. Why should we trust it? It has built-in biases through the information and data it uses. We know this, we need ways to overcome the biases, but we also need to understand how and why these systems arrive at their results. That's why we need people, otherwise bad things can happen," Prof. Tarjan argued.