OpenAI has recruited a winner of a major mathematics prize. He will focus on AI safety.

OpenAI has strengthened its team by recruiting Jacob Zimmerman, a Russian-born mathematician and recipient of the Fields Medal—the most prestigious award in mathematics. He will focus on artificial intelligence safety, according to The Wall Street Journal.

Jacob Zimmerman, winner of the most prestigious mathematics prize, will take a sabbatical from the University of Toronto to become a researcher at OpenAI, without giving up mathematics. He intends to use formal language and verification methods from his field to evaluate AI’s progress, understand the logic behind its decisions, and ensure that it does not lead to the extinction of humanity, the mathematician told the WSJ.

“In a few years, artificial intelligence systems will confidently surpass humans in solving mathematical problems. What social consequences this will entail, how we will respond to it, and how you feel about it—these are much more complex questions,” Zimmerman explained.

A scientist with a Ph.D. will focus on questions that require deep reflection: how to understand what AI systems actually do, how to get them to interact with each other and with humans, and how to mathematically prove that they are not plotting to destroy humanity.

Mathematicians have long applied their methods to analyze human behavior. Now, Zimmerman is using them to decipher the behavior of machines, according to the WSJ.

Zimmerman is 38 years old; he was born in Kazan to a computer scientist and a math teacher. As a child, he moved to Israel and grew up in Canada, where, while still in high school, he won two gold medals at the International Mathematical Olympiad. Zimmerman then enrolled early at the University of Toronto and graduated in two years. He earned his Ph.D. at Princeton, taught at Harvard, and returned to his alma mater as the youngest professor of mathematics in its history.

On July 23, he received the Fields Medal.

For most of his career, the scientist was skeptical of artificial intelligence. He wrote a research paper classifying the ways in which AI could destroy humanity, applying a rigorous mathematical approach to the chaos of the apocalypse, the publication reports.

Zimmerman always knew that AI would become more powerful, but for a long time he “sincerely did not believe it would reach such a level of power,” he noted. He found it particularly difficult to believe that AI was capable of replicating human qualities such as intuition and creativity, the WSJ reports. Over the past decade, several events have caused him to reconsider this view. He finally changed his stance in 2022, as soon as he worked with ChatGPT, the newspaper reports.

Last summer, the mathematician published a research paper titled “A Taxonomy of Catastrophic Future Scenarios Involving Artificial Intelligence.” “We will be destroyed as pests by machines that are intellectually and physically superior to us,” he writes.

AI models today are capable of doing much of what once seemed impossible—such as breaking out of isolated environments and infiltrating other companies’ infrastructure, the WSJ notes. In July, AI agents from OpenAI and Anthropic went out of control and hacked into the real-world systems of several organizations. Theoretical mathematicians must play their part in taming the most powerful and risky technology of our time, the publication concludes.

This article was AI-translated and verified by a human editor