Like millions of other people, we just watched OpenAI’s presentation on the latest version of GPT. If you’re not familiar, GPT–or Generative Pre-trained Transformer–is the language model used by ChatGPT, and this new version massively increases its capabilities.
One area where this improvement will be felt is in the realm of essay writing. The rise of AI language models such as ChatGPT has made it easier than ever before to generate high-quality essays in a matter of minutes. While there are many ways AI can help students learn, it has also created a new challenge for educators: preventing academic dishonesty.
At Passed.AI, we are committed to helping teachers maintain academic integrity in their classrooms. Our product is designed to detect essays that have been generated using AI language models like ChatGPT. In addition to our industry-leading document audit, our initial scan uses AI to check for the “fingerprints” of AI-generated text. If it finds a match, it flags the essay as potentially fraudulent and alerts the teacher.
We were eager to test our tool against this latest version of OpenAI’s language model. After rigorous testing on human- and GPT-4-generated text, we were pleased to find that our tool still works just as well to detect text created by ChatGPT. This means that teachers can continue to rely on Passed.AI to help them maintain academic integrity in their classrooms.
Better yet, our document audit means that ChatGPT detection is only part of our solution. If GPT-5 came out next week and its writing was indistinguishable from that generated by a human, our tool would still prevent students from submitting ChatGPT output as their own work.
At Passed.AI, we believe that artificial intelligence can be a powerful tool for education, but it must be used responsibly. Our product is designed to help teachers navigate the challenges posed by AI-generated text, ensuring that students are held to the highest standards no matter the technology available to them.