Ever asked an AI for risotto tips? Prof. Teamrat Ghezzehei has.
And while his mushroom risotto simmers on the stove, he’s also using AI to streamline climate change research, organize dense data scripts, and sharpen student learning.
Meet Teamrat Ghezze
hei, a soil physicist and full professor at UC Merced who investigates the physical basis of why soil is a habitable environment—especially for microbes. He teaches across Environmental Systems and Environmental Science and is just as at home exploring subsurface hydrology as he is testing a new photography AI.
“I started using AI about a year ago,” he recalled, “when my colleagues were making it do a bunch of crazy things. It was amazing at the beginning. I had never seen anything like that.” Like many, he was initially skeptical. “I didn’t take it seriously for about a month or two.”
But things changed. Drawing on his own experience with neural networks in soil physics, Teamrat began exploring Claude and ChatGPT more deeply. He prefers Claude, citing its ethical tone and the quality of scientific responses. “
From Risotto to Root Zones
Teamrat doesn’t just use Claude for research. He uses it to cook. “One day I wanted to make risotto,” he laughed. “But I didn’t have white wine. I had red wine, some dried mushrooms, onion, garlic… not very complicated.” He asked Claude for help. “It suggested I soak the mushrooms in hot water and mix that liquid with stock, and also to use a mix of red wine and apple cider. It actually worked pretty well!”
He’s also a passionate photographer and uses AI tools like DxO and Topaz to reduce noise in low-light images. “These tools are trained like GPTs, but on images. They go pixel by pixel to remove noise,” he explained. “One time I edited a photo of my son skateboarding. My niece—also into photography—said, ‘That face looks fake!’ She was right. Topaz had recreated it.” For Teamrat, this was an important reminder: “These tools are powerful, but it’s good to know how they work.”
A Bicycle for the Brain
When it comes to research, Teamrat uses AI to extend—not replace—his own expertise. He describes AI as a “bicycle on steroids for the brain,” referencing Steve Jobs' metaphor of the computer. “If you don’t know your stuff, they’re not going to be that useful. But if you do, they allow you to do many, many things much faster.”
For example, he used Claude to improve a dataset analysis on evapotranspiration trends across California. “I had the R scripts and the data,” he said, “and I asked Claude to help streamline the code and suggest better statistical techniques.” Claude introduced him to Sen’s Slope, a method Teamrat hadn’t used before. “I ended up rewriting the whole code set using that. 80% of it was AI-generated, with my instructions. I turned it into a poster for AGU this past December, and it looks like this is shaping up to be a paper.”
Teaching Students How to Think, Not Just Prompt
Teamrat is also exploring how to teach students to use AI effectively—not just as shortcut engines, but as tools for critical thinking. In one assignment, he asked undergraduates to read a scientific paper and compare their own analysis with what an AI tool said. “I wanted them to learn that reading is still important. But I don’t think I was totally successful,” he admitted. “These tools are getting incredibly good at analyzing text.”
Still, he allows AI use in writing assignments—as long as students aren’t substituting it for their own thinking. “If you're submitting a report that’s full of spelling and grammar mistakes, that’s not acceptable. Use the tools to polish. But don’t just ask AI the exact question I asked you and turn in the answer.”
A Note from Me:
Teamrat’s interview reminds me that AI use among faculty isn’t monolithic. It’s thoughtful, varied, and often surprisingly personal. Whether he’s tweaking code for climate research or crowdsourcing risotto tips from Claude, Teamrat shows how AI can serve as both a creative and analytical partner—if you know your science. His approach makes one thing clear: AI isn’t a substitute for expertise—it’s a multiplier of it.Want to share your AI story?Email sghosh@ucmerced.edu
