At 88, the father of AI worries about his grandchildren

• Published date: 09 September 2024
• Author: Assaf Gilead and Alon Pearl
• Publication title: Globes (Rishon LeZion, Israel)

But the truth, we realized, mainly after we parted, is very different. Pearl is a professor of computer science and statistics at the University of California at Los Angeles (UCLA). His résumé includes dozens of prestigious awards, and he is considered one of the fathers of artificial intelligence, who decades ago already saw the world we are heading towards. In addition, his life is marked by one tragedy that cannot be forgotten - his son, the Jewish-American journalist Daniel Pearl, was kidnapped in 2002 in Pakistan and murdered by his captors

Our interview takes place at 11:00 pm in Los Angeles. One would think that Pearl, who celebrated his 88th birthday last week, would want to end the conversation and go to bed. But he is more alert than ever, as if he had just started his day. In general, despite everything he has been through and accomplished, he doesn't seem to have any intention of stopping anytime soon.

"There's a lot more I want to achieve," he says. "I have so many ideas in artificial intelligence. I know there are things I can do and others don't know. Also, on a societal level, I feel I have a lot to offer the Jewish world. We don't know how to defend ourselves, and express Israel's righteousness in a way that can overcome leftist opinions. I feel I can do it."

"Artificial intelligence frightens me"

When we ask Pearl about the future of artificial intelligence, it is evident that he is in a state of constant dissonance. On the one hand, this is what he works on, this is what he lives for. On the other hand, he has many concerns. "Artificial intelligence frightens me," he says. "The creature we are building now may have revolutionary ideas and it may be the Golem that turns on its creator. As a scientist, I don't spend much time on these thoughts, but perhaps I am sinning against my grandchildren if I continue working on it, instead of stopping."

To better understand his concerns, one must dive a bit into Pearl's life's work. It's hard to believe, but his American dream actually got its start in Israel. He was born and raised in Bnei Brak (we'll get to that later), and in the late 1950s began his professional career studying electrical engineering at Technion - Israel Institute of Technology. The move to the US followed; Pearl completed a master's degree in physics at Rutgers University, and a doctorate in electrical engineering at the Polytechnic Institute of Brooklyn. In 1970, he joined the UCLA School of Engineering and Applied Science, where he still serves as a professor of computer science and statistics, and as director of the Cognitive Systems Laboratory at UCLA.

In California he entered a newly dawning era: the golden age of artificial intelligence, of which, in time, he himself became one of the mainstays. During those years, the academy and the commercial market were focused on developing decision-making AI with expertise comparable to that of professionals like doctors, academic researchers, and oil prospectors. But recreating the expertise of human professionals almost never worked, as they encountered a high level of uncertainty.

This is where Pearl entered the picture. He developed the theory of Bayesian Networks, based on the following principle: to identify a certain disease it is not necessary to understand the relationship between the various symptoms, but rather, the degree of their presence in the patient's body and the degree of probability for each, so that it is possible to identify their combined appearance as, for example, malaria or AIDS. This theory, some 50 years later, won him the most prestigious award in computer science, the Turing Award.

The article, published by Pearl in 1982, laid foundations for a new era in artificial intelligence that have lasted for about 40 years. He also turned his theory into one of the fundamental principles of artificial intelligence in the 1980s, 1990s, and early 2000s; from medical analysis and bioinformatics, to classification of email, Bayesian Networks allow many disciplines to structure their data and knowledge.

But Pearl wasn't satisfied with that. Later, he generalized the theory so that it would not only...