Edward Tian was in deep sleep when his bot crashed a website.
The 22-year-old senior at Princeton spent the winter holidays at his local cafe creating GPTZero, an app that he claims can tell “quickly and efficiently” whether an article was written by a human or by OpenAI’s ChatGPT. When he uploaded this to his app creation and hosting platform Streamlit, he didn’t expect it to be achieved. he is Attention
“I expected at most a few dozen people to try the app,” Tian told The Daily Beast. “Suddenly, the user count was insane, with over 2,000 people signing up for the beta within a few hours.”
GPTZero eventually saw such a huge influx of users that it even crashed the platform that hosted it. “I’m surprised it went viral,” he added.
When OpenAI released ChatGPT on November 30, 2022, it released a digital Pandora’s Box to the world.
From high school teachers to college professors and journalists, everyone feared the powerful AI chatbot, which some called “AIgiarism,” ushering in a new era of bot-generated essays and articles. Some educators are already reporting that their students are using ChatGPT to create end-to-end essays and finish writing assignments.
There’s still no official method of doing this, although OpenAI says they plan to eventually “watermark” something to verify if something was created by ChatGPT; journalism.
Tian, a double major in computer science and journalism, was troubled by the ethical dilemmas posed by chatbots, as well as the “black box” nature of major language models like ChatGPT. The opaque nature of models causes people to fundamentally misunderstand and therefore abuse them.
So, despite being on the verge of graduating, he decided to spend the winter holidays developing a tool that could help people find out if a post was written by a robot.
“People deserve to know that writing is not human,” Tian sid said. “There has been so much excitement lately about ChatGPT and the AI generation that people deserve to know the truth.”
GPTZero uses two different metrics to assess whether a text was penned by a bot: surprise and explosion. Texts placed in the application will be assigned a number for both measurements. If the number is low, it is more likely to be created by a bot.
OpenAI’s Impressive New Chatbot Is Not Immune to Racism
Confusion is a measure of randomness in a sentence. If a sentence is constructed or uses words in a way that confuses the app, it scores higher for surprise. Tian said he uses the free and open source GPT-2 to help train his app for this metric.
Burst quality comprehensive Randomness for all sentences in a text. For example, human writing tends to have sentences of varying complexity. Some are simple. Some might give James Joyce a shot for his money. Bots, on the other hand, tend to generate sentences of relatively low complexity throughout the entire text.
“There are beautiful qualities of human written prose that computers can never use and choose from with,” explained Tian. As a journalism student, he was inspired by a course he took with American writer John McPhee, who taught him these beautiful qualities of human writing.
Tian would go on to use an article written by McPhee. New Yorker As part of the GPTZero demo:
Despite improving the tool, Tian is not anti-AI. He believes there is a time and place for them if used ethically and with consent. Hell even used artificial intelligence programs like CoPilot to “support most of my coding”.
“I am not against using AI to write when it makes sense,” he said. “But people deserve to know that the writing is not human. There has been so much excitement lately about ChatGPT and the AI generation that people deserve to know the truth.”
With the hype and fears surrounding ChatGPT, a tool like Tian’s can be incredibly useful in all industries, from educators who want to see if their students have stolen an article, to recruiters who want to check whether a cover letter was actually written by an applicant. . As such, it can be incredibly lucrative for the right investors, some of whom have already reached Tian with dollar signs in their eyes.
“A bunch of VCs popped into my Twitter DMs the other day,” said Tian, including the likes of A16Z, Menlo Ventures, and Red Swan. But he’s not done with GPTZero yet. It wants to further refine and develop the app and even plans to expand its transparency with “explanatory and detection methodologies”.
And at the end of the day, he’s a senior in college. They have homework to worry about and finals coming up with man-made trials. Right now this is a much bigger concern than a digital Pandora’s Box or VC investors.
“I will answer all calls for now,” he said with a laugh. “But I’m just a college student focused on graduating from school.
Read more at The Daily Beast.
Got a tip? Submit to the Daily Beast here
Get the Daily Beast’s biggest news and scandals straight to your inbox. Sign up now.
Stay informed and get unlimited access to the Daily Beast’s unique reporting. Subscribe now.