NYC NOW - Columbia University Suspended Student Who Built AI Cheating App, so He Dropped Out and Raised $5.3 Million for Startup

Episode Date: May 10, 2025

Roy Lee was suspended from Columbia University after creating an AI app that helped users cheat on technical interviews. He dropped out, went viral, and raised $5.3 million to grow a startup. WNYC’s Janae Pierre speaks with reporter Ryan Kailath about what the story reveals about AI, ethics, and academic accountability

I'm Roy. I kicked out of Columbia University for building a tool to cheat on Lee Code Style Technical interviews, and now I'm building a tool to cheat on everything using AI. From WMYC, this is NYC Now. I'm Jenae Pierre. Happy Saturday. Quick question for you. If you build an app to help people get ahead, does that make you a visionary or a cheater? WMYC's Ryan Kylath recently reported on a Columbia University student who was suspended after creating an AI-powered app that helps users ace job interviews.

And now, Silicon Valley investors have poured millions into the app and its creator, Roy Lee. I literally don't know a single person at Columbia, at least in the undergraduate school, who has not used AI on an assignment before. What you think is deception today is just going to be normal tomorrow, and the sooner you embrace it, the sooner society can adapt and evolve.

Now, Ryan talked with him about why his invention sparked such different reactions and what that tension reveals about how colleges handle innovation when the rules haven't quite caught up. All right, Ryan, tell me the truth here. Did you use AI for this story? I did. I'll tell you about what I did at the end. Okay, yeah, we'll save that for later.

What can you tell me about Columbia student Roy and his friend slash classmates? Who are they? Yeah, so Roy transferred into Columbia as a sophomore last fall and knew that when he got to school, what he really wanted was to find a co-founder and a wife. He said, this is the value of an Ivy League. So he immediately finds a guy who shares his vision, a junior in the engineering school named Neil Shunmugam, and they decide we're going to build a company. So in the tech world where these guys work, there's a very normalized thing in the interview process where you do all these coding tests live in front of your interviewer because they

want to see how you think. How would you code this little thing? It's almost like an SAT or another standardized test. Roy said he had spent 600 hours training for these in order to be able to do these interviews. But then he thought, oh, what if we build a company that helps you cheat on these tests with AI?

He wouldn't really call it cheating. He would just say making the system more efficient. But that's what he did. Roy built this tool that sits undetectable on your computer while you're in a job interview and lets you use AI on the coding problems. But his real

innovation was, okay, I've built this cool tool that a lot of people in the tech world will want to use, but it's not going to work unless we can go viral with it. Guys, this is Roy from InterviewCoder. Today I'll be showing you how it works on a real Amazon OA. So he filmed himself using his tool to land internships at all the top companies. Wow. He got offers from all of them. He posted it and blew up. All right. That's pretty clever. So tell me, why was Roy's suspicions? Yeah, so Amazon in particular did not love this. They sent a strongly worded letter to Columbia saying, hey, your student just did this. You and me, Amazon and Columbia, we have a strong ongoing relationship between our company and your engineering school.

Would be a shame to see something happen to that. Columbia disciplined him. He filmed the disciplinary hearing and posted that too. Hey, capture lightning in that bottle. And so they kind of suspended him on the technicality. They were like, you weren't allowed to film and post that disciplinary hearing, so now you're suspended. So Roy has talked about how normalized the use of AI is already on campus at Columbia. What did he tell you about that?

Yeah, in his words, he didn't know a single undergrad at Columbia who does not use AI in an unauthorized way on their assignments. It's just like so broadly accepted that you'll use AI to write your essays or to do your assignments. And genuinely, I would not be surprised if it was something like 99 to even 100% of undergraduates use AI when they're not allowed. I was a student at Columbia in the grad school last year and I saw pretty much the same thing. I'd guess probably half the students were using AI on their assignments in a way that they're not supposed to. And by the way, we forgot to mention that the name of this app is Cluelly. But the tagline is cheat on everything. Now, Ryan, that's obviously going to raise some eyebrows, right?

Yeah, and this is part of his marketing stunt. He told me, you know, that tagline is more marketing than mission. She done everything is sort of intentionally ambiguous. It's just meant to be provocative. But I think if you took a few seconds to think about it, then you would realize that, one, the future where we use AI more and more and more is sort of inevitable. And two, as society adopts this future, everyone is going to be uncomfortable with it. But it's better that we just embrace the.

discomfort and sort of stand in the eye of the hurricane when it happens. So the product isn't just for these tech interviews anymore. He's envisioned Clue Lee as an AI layer that sits on your computer for every kind of virtual interaction. But he's saying it's not really cheating. It's just research. He used the example of a calculator. Back when the calculator came out, people would go on marches to stop people from using the

calculators. And what do you know, 20 years later, everyone literally has a calculator in their back pocket. AI use is just going to become more widespread. Right. Look at us now. Look at us now. Yeah. Even spell check, right?

Exactly. It's been a minute for me since I've been in college, didn't do the grad school thing because, you know, know thyself. But I'm curious, what about professors is their stance on AI? Most schools that I've looked at, they kind of leave it up to the professor on a case-by-case basis. At Columbia, you're allowed to use AI if your professor gives you explicit permission. And this is where you get into an interesting. thing with you've got a lot of younger, more tech savvy professors who are saying, okay, AI is here to stay.

How can I build it into the classroom? How can I build it into assignments? And how can I challenge students to show me the value that they, the human being, are adding on top of AI? At the same time, you've got some other professors who maybe they're near the end of the career. They've been teaching Shakespeare the same way with the same tried and true assignments for 40 years, and that's not going to work anymore because the students can use AI to diagram

the Shakespeare sentence or pull out the themes from the reading or write their essays. So I spoke to one professor who falls more on the progressive side of that. Onondrao, he's the chair of the communications department at the University of Mary Washington, also my alma mater, and he was my professor. He had an interesting take on this. He said AI is useful in a very specific way for them. looking at a disruption that brings back the need for those humanities and liberal art skills in a way that for the last couple of decades, people have kind of shirked.

And they've said that those aren't as important anymore. You need technical skills. Well, now it's a little different. Those leadership skills, the communication, collaborative, critical thinking skills are incredibly important, especially if you're going to be leading AI agents and tools in the workplace. So that has kind of framed my approach. He said that honor policies and cheating policies are going to have to, evolve and adapt because our definition of what cheating is is changing. What does it mean to use a tool

appropriately or inappropriately? Is it matter of just letting people know how you're using the tool? Maybe that's enough. How do you negotiate between those things? He's kind of at the forefront of this. He wrote a book on the subject of AI in education. So I think we're going to see everybody slowly catch up to the conversation. What kind of message does that send to students building things like this. So I think you're always going to see ambitious tech kids who have that. But Columbia also attracts a lot of more by the book, traditional, I'm here to go through school and then get a nice job in my field. But the entrepreneurial folks, they often have this kind of rebellious streak. So I think students who have that are going to admire him. Students who don't are going to

think this is a pretty wild gamble. He has already dropped out, right? Yeah, exactly. So he was suspended for a year and took the opportunity to drop out. All the publicity let him raise $5.3 million in a seed round, which is pretty healthy for a company with average age 21 and, you know, two people. He moved to San Francisco, got a loft. They're out there coding into the middle of the night. You know, it's just like the social network, basically. I can see Roy Lee right now in his San Francisco loft ordering to. Chinese takeout, coding until the night. You know, after all of this, though, what's your takeaway

from Roy's story? I come down, I think, on the side of this is the future that's here and it's time to adapt to it. And it'll be interesting to see how institutions adapt to it slowly, quickly, or not. Yeah. On that note, you mentioned at the top that you may have used AI for this story. Tell me more about that. Yeah. So I got my editor's permission. I just said, why don't I use Roy's tool in the interview to see if I can learn anything or improve anything. Now, his tool, it would have been too much just to download and install or whatever. So I did just pull up chat, GPT. And I said, like, these are the questions I'm thinking about asking him.

Do you suggest anything else? Or are there things I haven't thought of that would be really interesting to ask? And unfortunately, it didn't give me anything I could use. All right. Well, you still have a job. It's still adding value. That's WNYC's Ryan Kailat. Thanks, Ryan.

Thanks. Thanks for listening to NYC Now. I'm Jene Pierre. We'll be back on Monday.