GPTZero creator Edward Tian speaks with Yahoo Finance Live about his creation that detects when ChatGPT technology is being used and why he still sees a use for AI within education.
- Open AI's ChatGPT is smart enough to pass a final MBA exam at the Wharton School. That's according to one of the University's professors. So how does higher education know it's a student or a computer program submitting their next paper? Enter GPTZero.
The new program, created by a 22-year-old Princeton student, allows users to copy and paste text and test whether it was computer generated. GPTZero tests the texts and provides a perplexity score followed by a GPTZero score. That zero score determines whether the text is human generated.
In the result you're watching now, "The New Yorker" article shown was, in fact, written by a human. Joining us now is the creator of GPTZero Edward Tian. Edward, great to see you. Bravo. How do this is absolutely foolproof, and do you get a sense of how prevalent cheating is with ChatGPT?
EDWARD TIAN: First, thank you so much for having me. It's a pleasure to be here. I would say, yeah, GPTZero was something I built over my holiday break just at my local coffee shop in Toronto. And I was completely blown away at how it just literally went viral. It is not foolproof.
Yeah, we put a disclaimer not to have any make definitive academic decisions off the beta, but what we have done is put up a product wait list for educators that are interested in using it professionally. And so far, over 23,000 educators from more than 40 states and 30 countries have signed up for that. So we're really excited.
SEANA SMITH: Edward, you make it sound so easy. I think if I was tasked with this, I don't know if I could do this over an entire lifetime. But what was your motivation for this? Because I know you're writing your thesis about AI and you have a lot of interest just about AI and the role it will potentially play in the future.
EDWARD TIAN: Yeah, for sure. Aside from the research and academic interest, I think the concern was the immediate need in terms of education to sort of adopt this technology responsibly because I'm not against students using ChatGPT. Instead, I'm actually for students should and our generation should be exposed to these new technologies, but it has to be done responsibly and fairly, but as well as bigger concerns in terms of bots I've looked at in the past, like over at internship with the BBC, like bots, Twitter disinformation, as well as fake news being generated by these texts. That's all of concern. And we need to have tools and safeguards so these technologies are used and not abused.
- What are you hearing from teachers? What are you hearing from students that gives you a sense of, again, how prevalent cheating is with this technology?
EDWARD TIAN: I would say in terms of from teachers the reception has been phenomenal. In fact, it was just back at my high school this Friday and talking to all my teachers and my high school English teacher was telling me that, yes, students-- like they have no doubt that their students are all using this. But what's exciting is some teachers are thinking about also integrating in their curriculum in terms of maybe asking it to help generate writing topics or prompts.
The jury's still up in how educators are adopting these technologies because these technologies are here. We can't ignore that. And we have to build the tools to adopt it responsibly.
SEANA SMITH: Edward, talking about exactly how this works when it comes to detecting cheating or if someone if a human, in fact, wrote this, what are the certain things that you're looking for in order to alert that teacher, that professor, that this has actually been plagiarized?
EDWARD TIAN: Yeah, one of the indicators that GPTZero is looking for is burstiness. Burstiness, in a sense, is variance in writing in terms of-- you can think of it that humans have creativity and because our short-term memory we have sudden bursts and creativity and differences in our writing, while machines have pretty ubiquitous and constant writing over time, especially if these machines are as powerful as ChatGPT.
- The New York Public Schools recently banned use of ChatGPT. Do you think districts around the country should follow suit?
EDWARD TIAN: No. I actually don't believe that's the right approach at all. First of all, if I was a high school student, I'd just be using ChatGPT on my own public Wi-Fi or Wi-Fi at home, instead of at schools.
Secondly, these technologies are here. And I absolutely believe AI is here to stay. It is the future. And students shouldn't be taken away the opportunity to interact. It should be adopted responsibly, as well as soon ChatGPT will cost money.
So it's no longer just a responsible AI issue, but also an equity issue because students in low income neighborhoods might just never have access to this technology, when students in higher income neighborhoods might be paying for it. But if there's a blanket ban, then we actually don't know where it's being used as well.
SEANA SMITH: Yeah, certainly many factors to consider here. Edward, when it comes to GPTZero, I know it's only been several weeks here, but I'm sure you've gotten a tremendous amount of interest from VCs, maybe some larger firms that are showing interest in you and also your product. Have you gotten that interest, and what have those conversations been like?
EDWARD TIAN: Yeah, it's actually super exciting. Yeah, those from media being a journalism student, this is like really exciting for me to talk to people, as well as being able to talk to VCs. I would say at the moment we're focused on the engineering and actually building this tool out. And soon, yeah, we're happy to think more on the business side.
In terms of individual educators, how this looks like is that I'm pretty committed to keeping GPTZero or at least the copy and paste and enter online version free forever. We do want to have a tool that everybody can access because it is an important tool to have. And I think everyone deserves a tool like GPTZero to use as well.
- Edward, what do you think the best practical application for this technology is? We've heard theories all over the map, but you've studied it. Where do you think we'll see it develop the best in our economy and our society in the next couple of years?
EDWARD TIAN: Yeah, I think like AI generation is going to be everywhere. In fact, I actually use it in my own coding. I use co pilot which is GPT for co generation, just like ChatGPT is GPT for chat generation.
So it's really going to be ubiquitous. I know Microsoft is already looking at implementing OpenAI models into their products as well. So it's really going to be more and more accessible.
SEANA SMITH: Edward, in terms of the advantage that Microsoft now potentially has because of the investment with OpenAI and ChatGPT, is there anything out there from your experience from the research that you've done that's even close to where ChatGBT is today?
EDWARD TIAN: Yeah, I would say in terms of research as well, there's many, many other models that are doing those work, not just like OpenAI models, but there's models in different countries. I know Tsinghua University has lots of research and natural language models, as well as BERT-based models. as opposed to these GPT-based models, but I really don't think anything has broken the barrier in terms of accessibility like ChatGPT has. No other group really comes close in terms of how popular ChatGPT has been used.
- Just a quick question. Any of your Princeton classmates furious at you for developing this product?
EDWARD TIAN: I wouldn't say so. I think it's been pretty awesome. My classmates have all been pretty supportive. Professors as well have been pretty supportive, including my own thesis advisor. They're all pretty happy and proud.
- All right, I've got a 15-year-old at home who's not so happy with you. Edward Tian, congratulations, good stuff. Great to have you.
EDWARD TIAN: Thank you so much for having me.