Students using technology to cheat on essays is a tale as old as time. However, some educators are so concerned with ChatGPT that they've banned it in schools! That problem may already be old news Here's the story of how a Princeton student is combating AI plagiarism with AI🧵
When @OpenAI released their language model ChatGPT-3 back in November, it was clear that this was a step-change technological improvement. Naturally people began to think of new ways to apply this model. Before long, the technology had made its way into the hands of students...
Students have long leveraged the internet/technology to get out of writing essays. Whether it was using Sparknotes, changing text to white, or hiring people online, students will do literally ANYTHING to avoid writing 😂 And ChatGPT appears to be the best option ever.
With one simple prompt, students can have ChatGPT write entire essays. These essays are grammatically & informationally correct, and ORIGINAL pieces of content. This makes plagiarism much more difficult to detect since it's not a copy/paste from somewhere else on the internet.
Given the power of this tool, many educators took to social media to begin voicing their concerns. The fear was so great, that the NYC public school system even went so far as to ban ChatGPT completely from their school networks and devices 🤯 https://twitter.com/business/status/1611471144783060992?s=20&t=wvTzIQRX99ZuiBqN3KonJw
But it appears that those fears might have been overblown, as Princeton student @edward_the6 believes he may already have a solution. The best part is that he is combating AI-based plagiarism with an AI tool of his own called "ChatGPT Zero". Here's how it works...
The app relies on 2 primary indicators to detect if a blurb was produced by ChatGPT: 1. "Perplexity": How complex is the text. 2. "Burstiness": What is the variation between sentences (i.e. short sentences alongside longer ones). https://twitter.com/edward_the6/status/1610067688449007618?s=20&t=7NmMp4wFgQNKPuJEx5lxgw
Both of these indicators hone in on aspects that are unique to the ways that humans write. The human mind still tends to produce more naturally complex pieces than AI chat bots. And they are also more prone to variations in their writing style, unlike a chat bot which is rigid.
The Beta is live for anyone to try and so far over 20,000 people have already tested it. While Tian admits that the app is not foolproof, it still represents a step in the right direction to combat AI plagiarism. http://gptzero.me/
I think there is an even more important takeaway from this story: We should EMBRACE new technology not fight it. As AI continues to improve, I am expecting a wave of hate and resistance to pop up, as has always been the case throughout history with new technologies.
This resistance dates back to the 19th century, when a group of English textile workers known as the 'Luddites,' went out and destroyed textile machinery. They believed that this new technology was ruining their lives and that the only way forward was by eliminating it.
However, as we saw time after time, these technological breakthroughs led to NEW jobs & MORE prosperity for society in the long term. While it may lead to short term disruption & discomfort, our natural reaction should be to leverage new technology like AI for good vs hating it.
The NYC public school system is acting like Luddites, whereas Edward Tian is acting like an innovator. If I had to place bets for which approach will have more success in the future, I'm definitely betting on the latter!