GPTZero and AI Text Classifier are here to Combat ChatGPT Misuse
A 22-year-old student develops GPTZero, and OpenAI itself releases AI Text Classifier to curb the misuse of ChatGPT and AI text in academics and disinformation.
GPTZero and AI Text Classifier
If you are a tech enthusiast or internet user, you must have heard about and used ChatGPT. If you still need to, read our guide to ChatGPT.
It’s an interactive chatbot powered by machine learning created by AI startup OpenAI. It has devoured the entire Internet, reading the collective works of humanity and learning patterns in language that it can recreate. Its buzz has always been up since its release in November 2022.
Everyone is awed by its ability to perform an endless array of tasks just by typing simple commands in its dialogue box. This could include writing a story in a particular style, answer a question, explaining a concept, composing an email, or writing a college essay.
Although amazed by this ChatGPT, for Edward Tian, 22, a Computer Science graduate from Princeton University, like many others, wonderment quickly turned to alarm. As a student, he was apprehensive about its impact on the education system.
Worried about the increasing use of generative AI, an article published in The Atlantic declares that college essays are dead as students use AI instead of their creativity.
Edward Tian Creates GPTZero
According to the original story published in NPR, when everyone was partying on New Year’s day, Edward Tian was busy finding solutions to combat the misuse of ChatGPT. The idea that hit his mind was to apply what he had learned at the university to create a tool to identify whether a human or machine had written something.
So, by using GitHub Co-Pilot, which is also powered by GPT-3—an AI system also used by ChatGPT—Tian was able to develop an app within three days.
He released his app, GPTZero, on January 2nd, which uses ChatGPT against itself to check whether “there’s zero involvement or a lot of involvement” of the AI system in creating a given text.
Initially, he didn’t expect much from GPTZero. But when he woke up the next morning, his phone was flooded with messages and DMs “from places as far away as France and Switzerland.”
According to the story:
“His app, which is hosted by a free platform, became so popular it crashed. Excited by the popularity and purpose of his app, the hosting platform has since granted Tian the resources needed to scale the app’s services to a mass audience.”
During the testing, we asked ChatGPT what Bitcoin is. Its response was:
“Bitcoin is a decentralized digital currency, without a central bank or single administrator, that can be sent from user to user on the peer-to-peer bitcoin network without the need for intermediaries.
It was created in 2009 by an unknown person using the pseudonym Satoshi Nakamoto. Transactions are verified by network nodes through cryptography and recorded in a public distributed ledger called a blockchain. Bitcoin is unique in that there are a finite number of them: 21 million.”
When pasted in GPTZero, it gave the following result:
“Your text is likely to be written entirely by AI.”
Clearly, the app detected the involvement of AI in our tested text. You can give Tian’s GPTZero a try here.
OpenAI also Releases AI Text Classifier
To curb the misuse of ChatGPT, its developer OpenAI has also released a tool to distinguish between text written by a human and text written by AI from various providers.
The tool, dubbed AI Text Classifier, is a “fine-tuned GPT model that predicts how likely it is that a piece of text was generated by AI from a variety of sources, such as ChatGPT.” You can give it a try here.
Like ChatGPT, AI Text Classifier has limitations, such as unreliability below 1,000 characters, sometimes mislabeling human-created text as AI-written and vice versa. AI-written text can be edited to evade the classifier.
However, it is very satisfying that AI misuse is not going unchecked. Work is being done to complement the undeniable benefits of AI.