How Searching for the Higgs Prepared this Physicist to be an AI Leader in the Corporate World
For Sarah Schlobohm, a physics degree led naturally to machine learning.
A PhD in particle physics should come with a free programming degree, says Sarah Schlobohm.
As a graduate student working on Fermilab’s DZero experiment, a precursor to the ATLAS and CMS experiments that discovered the Higgs boson, Schlobohm spent many long days running algorithms on complicated data. At first, she didn’t know she was using machine learning — she was just doing research. Now she spends her days using machine learning and artificial intelligence (AI) to address challenges in the corporate world.
The move from analyzing particle interactions to business transactions wasn’t much of a stretch for Schlobohm. “You’re pointed at some slightly different data, but a lot of the techniques, a lot of the processes, are still the same,” she says. That includes specific models and algorithms and a scientific approach to solving problems.
But it wasn’t the route she initially expected to take. As a new doctoral student, Schlobohm imagined herself becoming a physics professor — the stereotypical path. But as her thesis neared completion, she realized that there were far from enough academic positions for all the graduate students who wanted one. So, like many of her peers in particle physics, she started exploring jobs in data science.
“I was nervous about leaving science,” Schlobohm recalls. “Would I be able to take this [physics education] anywhere?” The answer, she found, was a resounding yes. The skills and techniques students learn throughout a physics education are transferable — and valued — outside of academia.
Now, she’s all about AI. Machine learning and AI have a lot to offer the business world, according to Schlobohm. And the opportunities have only grown with the advent of generative AI — which can create new content or code — and its newfound accessibility.
“We’re in this AI moment,” Schlobohm says. She recalls when the internet became a reality, and then Google. “That's what this feels like. We don't know what's going to happen yet, but it's going to be big.” Schlobohm expects to see many breakthroughs in the next few years, but it’s already clear that generative AI models can code, debug, and process information much faster than traditional methods. That makes it excellent for addressing efficiency problems, she says.
For example, imagine that a bank has detected fraud perpetrated by a business owner. Now it wants to know whether the fraud extends to associated businesses, and if so, how. Traditional investigative methods can be time-consuming, but using AI, investigators can map the fraudster’s relationships with people at linked companies and more quickly identify people of interest.
Much of Schlobohm’s career has involved using machine learning and AI in the financial sector: auditing financials models, assessing credit risk, detecting fraud, and using data science to prevent crimes like money laundering and terrorist financing.
Most recently, Schlobohm was head of AI at a global technology consulting company that works with businesses across sectors, from healthcare to green technology. Her next move is into the human resources space at a company called the Citation Group, where she’ll be working for — believe it or not — a PhD geophysicist.
“I work with a lot of physicists,” says Scholobohm. “I've hired some physicists, in part because I know what the training is.” That training includes problem solving and data analysis, but valuable soft skills, too, like the ability to ask good questions and communicate technical information to nonspecialists.
In generative AI, your output depends on the model, its training, and the question you ask. “You have to think deeply about what problem you’re trying to solve and how,” Schlobohm says. And asking the right questions is so important that it has its own buzzword in AI: prompt engineering.
“Good prompt engineering is how you get generative AI to give you a good answer,” Scholobohm says. “But that’s just another way of saying ‘ask a good question.’” She credits science with teaching her what ‘good’ means, and she appreciates that any physicist she hires has that knowledge.
Physics students often develop the second soft skill Scholobohm credits for her success — communication — during extracurricular science outreach activities. Outreach experience “is so valuable in a business setting,” she says. “That ability to translate the technical to people who are clever, but not technical, and especially not technical in that area, has been so important,” Scholobohm says.
The business world has a different culture than physics; you have to interact with clients, dress nicely, and be polite, according to Scholobohm — even when someone asks an off-the-wall question. “When someone comes up to you and says, ‘So does this mean we’re like, all really made of energy?’ You have to figure out how to answer that question accurately, but tactfully. It turns out that is super important in business and a lot of different areas.”
Schlobohm also regularly contributes to discussions about AI and ethics, giving talks about maintaining data integrity, overcoming biases in training data, and guarding against privacy leaks. She’s active in educating the next generation of data scientists by developing training modules, mentoring others, and talking at career events.
Behind it all is curiosity about how the world works. “I love thinking about new ideas,” Schlobohm says. “What if we take this idea and push it to its furthest possible limit?”