This past spring, when Harrison Shao (C ‘29) would get back to the Quad after a long day of classes, swim practice, Pottruck sessions, club meetings, or any of the normal exploits of an overcommitted freshman, there was one more thing he made sure to do: train artificial intelligence.
Whenever he got the chance, Harrison opened the AfterQuery Experts Slack channel and assumed his place in a massive, virtual assembly line manufacturing AI training data.
When we talked, Harrison had just wrapped up a project testing an AI model’s ability to read research papers on a variety of topics, from biology to finance. He had grilled AfterQuery’s client model on the experiments and results of over 50 biology papers, and he was part of a team processing what he supposed to be thousands of academic articles. Once he received the auto–generated answers to his questions, he graded the model’s response based on another worker’s “rubric” and sent the results to yet another team for review.
In addition to training AI, Harrison helped make marketing materials for AfterQuery Experts, the company’s contract worker–focused branch, allowing the startup to attract more AI trainers generating data in their fields of expertise.
Often, this marketing was aimed at his fellow undergraduates. AfterQuery, a startup that reached a $3.2 billion valuation after just eighteen months in business, has placed a heavy emphasis on recruiting college students since its inception. The startup has an especially strong presence at Penn—many of its contract workers and full–timers come from the university, as do two of its co–founders.
College students have always been eager to pick up part–time jobs and earn some pocket money, but this new kind of gig work is especially fitting for today’s hyper–online, AI–pilled economy.
Of course, the catch is glaringly obvious. College students aiming for entry–level white–collar jobs are extremely vulnerable to automation, and here they are, training AI to replace them.
Like many Penn students, Harrison learned about AfterQuery Experts from a peer: in this case, a friend at Duke.
“Over winter break, I just kind of scrolled through LinkedIn, looking at different startups,” he says. His friend had already assumed a part–time role at AfterQuery Experts, and Harrison came across one of his posts advertising a growth marketing position at Experts that paid an impressive $40 per hour.
The role seemed right up Harrison’s alley—in the fall, he had worked as a growth intern making Instagram reels and other user–generated content for AI startups hoping to grow their brands. Following the post’s instructions, he commented with his email and received a message inviting him to apply.
Harrison sent in his resume, answered a few short questions on the online portal, hit submit, and quickly forgot about his application as the holiday season picked up. A few months later, however, he was notified that his application was accepted, and he could begin onboarding.
“No interview,” he notes, despite the job’s generous pay. “I just got the offer, so it was surprising.”
Despite applying only to a growth analyst role, Harrison realized he could also earn some quick cash by training AI for the company. He soon became part of AfterQuery Expert’s sizable workforce—though it’s unclear how many users are active, the Experts Slack channel currently includes over 15,000 members.
AfterQuery’s rapid growth comes at a time when AI progress demands input from subject matter experts—a relatively recent development.
Before the AI boom, data annotation was a niche and relatively unattractive job. The first AI trainers might have spent their days categorizing movie reviews as positive or negative or labeling all the food and drinks in a refrigerator, often for single–digit wages. This annotated data, together with the vast troves of internet data OpenAI scraped to train their models, formed the basis for ChatGPT’s shocking fluency when it launched in 2022.
When training previous Large Language Model (LLM) generations, it was possible for lay annotators to judge whether an English response was coherent or not. But the ability to sound confident didn’t reflect true accuracy or expertise, leading to AI models’ infamous hallucinations.
For AI to do the work of real–life experts, they need to be trained by those experts. A doctor, for instance, could grade an AI model on its medical accuracy, comparing the AI–generated responses to her own until their answers gradually began to match. Done at a large enough scale, this sort of training improves an LLM’s accuracy by orders of magnitude across a variety of high–skill trades, from medicine to finance to poetry.
To attract working professionals, AI training companies often offer generous pay to so–called “domain experts.” These wages, buoyed by venture capital funding and a high demand for training data, have led thousands of consultants, software engineers, screenwriters, and researchers to put their skills at the service of LLMs.
Sometimes, these trainers are employees or freelancers supplementing their other income streams; sometimes, they’re laid–off workers training AI to do their old jobs.
In AfterQuery Experts’ case, the workforce also includes students attending high–ranking colleges, eager to do work related to their major and earn some easy cash.
At Experts, as in the rest of the data annotation industry, workers are staffed on projects with unrelated names—Egg, Red, or Super Red, for example—and perform one narrow task to help train an unnamed model.
“I'm not really sure which companies I work with, but I know they just have some huge contracts,” says Joanne Lin (W, E ‘29), a student who used to work for Experts.
In her role as an “undergraduate finance expert,” Joanne had two types of tasks: coming up with finance questions that would stump an AI model and reviewing others’ questions to ensure they made sense and fit the project’s needs.
For one project, Joanne grilled the client model on various financial documents from companies listed on the NASDAQ. She asked the model to perform complex retrieval and financial calculations, or forecast trends in the company. At the time, the project paid $40 a question, with a $1,000 bonus every 12 questions.
Putting together a good question was not always easy, especially given Joanne’s relative inexperience. “I only know very basic finance terms, and AI also knows basic finance terms,” she explains. “I guess it took around one to two hours for one question for me, which is why I, like, rage–quitted.”
When she worked for Experts, the young data annotator made most of her money reviewing other workers’ questions for $50 an hour, at a minimum rate of six questions per hour.
After onboarding in the spring, Harrison wondered if the recommended ten hours per week of work was a hard cap. “I asked, like, ‘Is ten hours the maximum?’” he recalls. “‘Can you go over?’”
In fact, there was no limit to the amount Harrison could earn, so long as he was staffed on a project. After spring break, he decided to take full advantage of this and devoted himself to the “dark grind.”
“Over break, I felt like I didn't really do anything. I didn't make any money,” he says. “So there was, like, four days in a row where I skipped all my classes … [and] worked eight hours every day.”
“I’m not gonna sit in [class] when I have more important things going on,” he continues. After making $1,300 in less than a week, it was hard to justify doing much else.
“I was actually gonna rush … a frat,” he explains, “but then I got the job, so I was like, ‘I need to make bank.’”
Like all good things, AfterQuery began as a Y Combinator–backed startup.
In 2025, founders Spencer Mateega (W, E ‘25), Danny Tang (W ‘25), and Carlos Georgescu left behind prestigious job offers in investment banking, Big Tech, and quantitative finance to found their own company in San Francisco. Once there, the trio worked out of their two–bedroom in Dogpatch and explored multiple AI automation–related ideas. Eventually, they realized current models weren’t trained to do the work of industry professionals.
Inspired by this very obstacle, they founded AfterQuery and secured a place in YC’s Winter 2025 class of startup founders. Their goal was to compile human–made datasets to train AI in industries like finance, software engineering, and law. The startup’s introductory post on the Y Combinator LinkedIn page announced that “if AI is one day to replace jobs entirely, it can’t just be good, it needs to be near perfect.”
AfterQuery did not respond to requests for comment on this article.
In short order, the co–founders began hiring project leads to procure contracts with frontier AI labs and oversee contract workers under AfterQuery Experts. By April, they had raised $50 million in a Series A funding round, with a total valuation of $500 million. Five months later, that valuation has more than sextupled, making AfterQuery the fastest–ever unicorn to come out of Y Combinator.
At the same time, undergraduates hired by Experts as growth associates and campus leads across top universities began recruiting other students as contract workers. This past spring, the Experts LinkedIn page announced roles for college software engineers and finance students. Applicants could reply with their emails to be considered—combined, the posts raked in almost 2,500 comments. Another LinkedIn post from a former Penn campus lead garnered over 2,200 responses from prospective AI trainers.
Perhaps because two of the three co–founders were Penn students, the startup has had an especially strong presence at the University. Advertisements for contract worker roles have appeared on the Wharton Investment Trading Group’s newsletter, and many current students—including the 2029 class president—have created annotated data or recruited classmates for the company.
While it’s hard to judge exactly how many Penn students and recent grads do contract work for the startup, Joanne believes the number is quite high. “I think a lot of them are Penn students,” she says. “I know I got a lot of my friends on there.”
When I talk to Pritika Kharkwal (W ‘29), an undergraduate finance student, she tells me that she used many of the same skills she learned from her part–time private equity internship to create finance training data for Experts.
“We create deliverables, and then kind of see if AI can do that,” Pritika explains. She might, for example, create an Excel sheet with sample data to calculate discounted cash flows—a staple task for entry–level finance workers.
“Experts will be asked to submit deliverables, and then … their goal is, like, bridge that gap [between humans and AI].” As she speaks, a shadow comes over her expression. “They're gonna train their model to perform to that capability … so you're kind of what they're measuring.”
After a short pause, she completes the thought. “So we’re basically becoming more replaceable, but, like, oh—I just realized.”
As we sit in the basement of the McClelland dining hall, she leans forward with her elbows on her knees and places her hands over her temples, staring blankly into space.
“This is gonna be a dark article you're writing,” she says.
The future of employment and AI is a sensitive subject for college students and fresh grads. Across every profession AfterQuery is training AI for—software engineering, finance, law—entry–level tasks are the most easily automated. Even as undergraduates learn the skills historically necessary to secure a job offer, some of them are passing along that knowledge to Claude, ChatGPT, and other frontier models.
“The hypothesis is … AI can currently do the types of things that recent college grads would do at an entry–level position,” explains Chris Callison–Burch, professor of computer and information science. The professor has become an authority on AI—in the fall, he teaches a class on the subject which draws in over 500 students. He says that the aspiring software engineers in his lecture hall occasionally express concerns about being automated away.
“I think computer science is on the front lines for this, because AI writing software has become incredibly good,” Callison–Burch says. “It's become much harder as a recent college grad to get your first job at a big technology company, and so I feel like the concern is grounded.”
As models extend their expertise to other white–collar industries, the professor fears that entry–level positions like paralegals “will go the way of the lamplighter.”
“There's lots of these studies on careers that are exposed to AI, and all of them are the types of things that we universities are preparing people for,” Callison–Burch says.
Indeed, the tasks that interns and fresh graduates perform, like note–taking, slide–making, document–compiling, or appointment–scheduling, are the most easily completed by an AI agent. Normally, this work justifies a junior employee’s existence while buying them time to advance. But as companies automate these tasks away, many are predicting an abysmal junior–level job market in the near future.
This is already the case for many of the most “AI–exposed” occupations. Take software engineering, in which employment of young workers (ages 22–25) is 19% below where it would have been if growth had kept pace with less–exposed sectors. Indeed, many companies are more than willing to hire fewer entry–level employees as AI systems improve—in a recent poll of over 600 recruiters, the Graduate Management Admission Council found that one–third of respondents are replacing entry–level jobs with AI.
Callison–Burch, who testified in front of Congress in 2023 on the “real possibility that generative AI may be able to replace a large number of white–collar jobs,” sometimes finds himself imploring company leaders not to pull that trigger.
“If an AI system is capable of making your employees twice as productive, you can make a decision,” he says. “Do you want your company to produce more, or do you want your company to have half as many employees?”
For now, at least, replacing workers with AI is probably an unwise choice. Many of the companies that announced layoffs earlier this year due to gains in model competency have reversed course after AI models proved unable to replace human experts. If companies like AfterQuery have their way, however, it may only be a matter of time before AI can do most human tasks.
After taking a beat to recover from her revelation, Pritika reasons as much. “I keep telling myself that AI can't give a presentation, AI cannot pitch to clients, but I think the more I see the world of AI, I realize there's a lot it can be taught to do.”
“Our life experiences, our memories, that's what makes up what we produce,” she continues, now glum. “The equivalent of that for AI would be data, and if we're giving it more data …”
If college students continue to train AI, it will only get harder for new students to act as effective trainers. As models improve, would–be associate consultants, junior analysts, software engineers, and paralegals won’t be able to stump LLMs or provide useful deliverables for training.
Around the same time, AI will be ready to replace college graduates, or at least change their roles so significantly that a wave of mass structural unemployment seems inevitable. To a stratum previously bound for the upper–middle class, the prospect is almost unimaginable.
“I came in like, ‘I want to be a software engineer,’ and then I wanted to be [an] investment banker [or in] consulting,” Harrison explains.
In the past, it was common for Penn students to weigh all these high–value desk jobs before committing to a single path, having unshakeable faith in their employability. But now, no matter what career Harrison considers, he realizes “it's going to be replaced by AI really soon.”
Rather than try to overcome these insurmountable barriers, some of the data annotators have chosen to accept this harsh reality. In other words: if you can’t beat them, join them.
Indeed, several of AfterQuery’s full–timers are past or present Penn students; many are picked out from the pool of contract workers and stay at the startup for at least a semester. Joanne’s sister, for example, wrote questions for Experts before becoming a strategic project lead at the company in her junior year. As Joanne recalls, “[my sister] would be sitting in front of a computer; she's like, ‘I need to make 20 questions so I can get $1,000 today.’”
“She’d grind those out every single day during winter break,” Joanne continues. “Maybe her work was really good or something … because one of the guys reached out to her.” Before long, Joanne’s sister reneged on her investment banking offer and joined AfterQuery as a project lead, overseeing her own teams of expert AI trainers.
Perhaps closer than anyone to the frontier of automation, many of these young model trainers conclude that AI fluency might just save them from the impending job apocalypse.
“I always like this quote,” says Harrison, “where it's like, ‘AI is not gonna replace us, but people who know how to use AI will.’”
