Matthew Simmons is a PhD candidate in literature at the University of California, Santa Cruz, whose dissertation research explores environmental writings rooted in California redwoods, climate, and landscapes.
His academic work asks questions that don’t resolve neatly. How do you represent something that outlasts you by thousands of years? How does language handle scale, time, and ecological systems that resist human-centered narratives? Those questions shaped years of reading, writing, and teaching.
As Matthew moved closer to completing his dissertation, a more practical problem surfaced. Academic jobs were shrinking. Summer funding disappeared. Teaching assistantships didn’t cover the full year. Like many PhD candidates, he found himself trying to finish deeply intellectual work while piecing together short-term income.
That was when he came across Mercor.
Finding Structure Through AI Training
Matthew applied to Mercor partly to practice interviewing after years inside academia, and partly because he needed something to bridge the gap.
Now at Mercor, Matthew works as a domain expert, contributing to AI training through prompt engineering and dataset refinement. His role involves writing questions that models struggle with, then carefully defining how those questions should be answered—what sources matter, what reasoning steps are required, and where errors tend to appear.
The work felt familiar in an unexpected way. Much of it resembled what he’d been doing for years as a humanities scholar: close reading, careful framing, and attention to how meaning is constructed.
“I create prompts that AI can’t answer,” he explains. “Then I teach it how to answer—how to find the information, analyze it, and come to the correct response.”
That process requires precision. Prompts can’t be vague, and they can’t be overly narrow. They need to surface real reasoning challenges while remaining evaluable. Matthew spends hours testing questions against multiple models, adjusting phrasing, and tightening constraints until the logic holds.
Working inside these systems sharpened his understanding of what AI handles cleanly and where it falters. Models perform well when questions have clear, well-documented answers. They need technical references, structured policies, and source-backed facts. Models start to struggle when questions require interpretation, competing perspectives, or contextual judgment.
“AI is trained to answer questions that have answers,” he says. “It doesn’t do nuance well.”
That insight guides how he writes training data. The goal is to expose complexity clearly enough that models can navigate it responsibly.
Finishing a Dissertation While Training AI
The practical impact of Mercor’s work mattered just as much as the intellectual fit.
Before Mercor, summers were spent searching for short-term work while trying to keep academic momentum. Once we completed a couple projects, Matthew settled into a steady rhythm. Mornings were reserved for dissertation writing. Afternoons were dedicated to AI training work.
“I wrote more of my dissertation in the last two or three months than I had in the previous year,” he says.
There was overlap between the two. On some days, Matthew built prompts around topics already central to his research—environmental change, information sourcing, and scientific uncertainty. He recognizes that AI prompt writing has sharpened his own writing. Structuring datasets reinforced habits of clarity and rigor.
He also found satisfaction in completion. Seeing a dataset finalized—hundreds of carefully reviewed prompts, edited and aligned—brought a sense of craft he recognized from scholarship.
“I like seeing a finished product that’s been meticulously put together,” he says.
When Matthew talks about what he hopes to contribute long-term, he returns to the same value that drew him to literature in the first place.
“I hope I can add some tolerance for nuance,” he says.
For him, AI works best as a starting point. It can help people orient themselves, find credible sources, and understand the boundaries of any single answer.
Explore opportunities for you at: www.mercor.com