‘My job is to worry’: Google DeepMind COO Lila Ibrahim on the responsibility of building AI
London
CNN
—
Lila Ibrahim's first love is not computers. Somewhat surprisingly for the first ever COO of Google DeepMind, it's not even artificial intelligence, nor is it coding.
Ibrahim's first love is engineering, and it's that background, she says, that makes her so valuable at a job that's all about computing.
"I became an [electrical] engineer because I thought it was a combination of math, art and science. And along the way, I actually really enjoyed working with people. And what I've enjoyed about my engineering career is the ability to bring all of that together and bring a unique view into everything that I do," Ibrahim told CNN's Anna Stewart during a recent interview at Google DeepMind headquarters in London.
"Being an engineer has taught me to ask the question of what, why, and what are we trying to achieve? So that if you can really understand a problem, you can figure out what the right solution is rather than just throw a bunch of solutions at an undefined problem."
As professional problem-solver, part of her job "is to worry," Ibrahim said. "What are the risks, and how do we mitigate them? And also to think about the opportunities, and how do we support them? … I feel like I had almost a moral calling to be in this role and all of a sudden, my very weird, circuitous background kind of makes sense with where I'm sitting right now."
One thing she has learned, Ibrahim said, is that she's not very good at predicting the future: "But I'm very good at building it."
Lila Ibrahim, pictured here giving a speech on stage in Geneva, Switzerland in 2023, has spoken openly about the need for careful consideration in the development of artificial intelligence.
Johannes Simon/Getty Images
Thinking from the outside-in
Growing up as the daughter of immigrants, with English as her second language, Ibrahim said she often felt "like an outsider" throughout her childhood and into adulthood, first in the American Midwest, then as an exchange student in Japan, and into university at Purdue, in Indiana, where she studied electrical engineering.
"There weren't that many women," Ibrahim recalled. "In fact, you could count them on a couple of fingers at the time."
By then, she added, in her early twenties, "I was so used to having to get comfortable with bringing a different perspective into everything."
Lila Ibrahim become the first COO of Google DeepMind in 2018.
Google DeepMind
While that "outsider" mentality initially felt like a hurdle, Ibrahim says the biggest lesson she's learned is to accept it as a superpower, and wishes she'd done so sooner.
She built her career, first at computer chipmaker Intel, then at a venture capitalist firm. Ibrahim went on to become the first president and COO for online learning platform Coursera.
In 2018, a fascinating opportunity crossed her desk: DeepMind, an artificial intelligence research lab founded in 2010, which Google acquired in 2014, was looking for its first COO.
WASHINGTON, DC - OCTOBER 10: Pepsi Chairman and CEO Indra Nooyi speaks onstage at the Fortune Most Powerful Women Summit - Day 2 on October 10, 2017 in Washington, DC. (Photo by Paul Morigi/Getty Images for Fortune)
Paul Morigi/Getty Images North America
Related article
‘In so many ways, I broke so many barriers:' Former Pepsi boss Indra Nooyi on life as a trailblazing CEO
"Thirty years into my career, I really wanted to be very deliberate and intentional on what this next chapter would be," Ibrahim said.
"But when you have a chance to work on such transformative technology … how do you say no? So I actually engaged in the conversations, but very slowly and intentionally. I wanted to understand, what were the founders' vision for what I could make possible, and what were the risks?"
In total, Ibrahim says she spent 50 hours interviewing for the position, as she weighed the prospect of entering the exciting but often controversial world of AI.
"I'd go home and I would tuck my daughters in at night saying, ‘what kind of legacy will I leave in the world?' At the end of the day … I felt there was no better place to build AI responsibly than DeepMind," she said.
A Nobel-winning idea
Ibrahim's love of engineering was inspired by her Lebanese father, who was orphaned by the time he was five years old and would grow up to become an electrical engineer.
"I remember [when I was] growing up, he would have these beautiful drawings on his desk at home," Ibrahim said. "And then I would see these pictures turn into microchips that would go into things like heart pacemakers. So this orphaned kid from Lebanon would actually have been able to save people's lives through his work on pacemakers, all through engineering."
Lila Ibrahim, pictured here with her father Shawki Ibrahim, credits him with inspiring her to become an electrical engineer.
Lila Ibrahim
Ibrahim's father was orphaned by the time he was five years old, in Lebanon.
Lila Ibrahim
Ibrahim (center) says she always felt like an outsider growing up in the American Midwest.
Lila Ibrahim
Motivated by her father's example, Ibrahim views her work in terms of impact. At Google DeepMind, perhaps the mostimportant example is AlphaFold, the company's AI program capable of solving what it calls the protein prediction problem.
"A protein is a basic building block of life," Ibrahim explained. "And if we can understand how a protein might fold, we can understand the function that it has and when it misfolds, what's gone wrong. And so things like Alzheimer's, Parkinson's, these are all protein-related problems."
What would normally take a human researcher years for just a single protein now happens in minutes. The company also made AlphaFold open-source, meaning any researcher anywhere in the world can access it (over 2 million people in 190 countries and counting, according to Google DeepMind). In October, two of Ibrahim's colleagues were awarded the Nobel Prize in Chemistry for the program.
"We were not expecting [the Nobel Prize]," Ibrahim said, "definitely not in this year," noting AlphaFold is only four years old.
"AlphaFold is just the first step," she added. "We have a portfolio of research that's happening not just in biology but in chemistry, in physics, and so much more."
