Turnitin has successfully developed an AI writing detector that, in its lab, identifies 97 percent of ChatGPT and GPT3 authored writing, with a very low, less than 1/100 false positive rate.
The company plans to add this functionality to its core writing integrity products as early as April 2023, according to a press statement. The new functionality will operate within the existing Turnitin workflow so that educators will be able to analyze content and use feedback tools in the same user experience they have today.
Turnitin also published an AI writing resource page to support educators with teaching resources and to report its progress in developing AI writing detection features. The newly launched AI writing resource page is publicly available and will be updated regularly with information about Turnitin’s progress in bringing detection features to market including how they are performing in its research and development lab. Turnitin experts in pedagogy and instruction will also contribute to an expanded library of resources to help guide K-12 teachers and higher education faculty on how to adjust to an academic environment where AI writing is used. Additionally, demo and preview videos will be regularly posted.
Speaking about the development, Annie Chechitelli, Chief Product Officer of Turnitin, said,
Based on how our detection technology is performing in our lab and with a significant number of test samples, we are confident that Turnitin’s AI writing detection capabilities will give educators information to help them decide how to best handle work that may have been influenced by AI writing tools. Equally important to our confidence in the technology is making the information usable and helpful and in a format that educators can use.
We are being very deliberate in releasing a detector that is highly accurate and trained on the largest dataset of academic writing. It is essential that our detector and any others limit false positives that may impact student engagement or motivation.
Adding further about the technology, Eric Wang, VP of AI for Turnitin, said,
We have developed state-of-the-art AI writing detection systems using recent advances in deep learning transformer architecture – the same technology that powers GPT-3 and other large language models (LLMs). Our AI writing preview has been trained on academic writing with high efficacy rates and can identify 97 percent of AI writing, with a very low 1/100 false positive rate. Launching it within our existing flagship products allows us to provide AI writing detection within the existing workflows that our customers are accustomed to, without having to copy/paste text or upload papers to a new, unverified third party.
Chris Caren, CEO of Turnitin, said,
For over 20 years, Turnitin has been a trusted partner of educators, giving insights into student writing, and helping to uphold the standards of academic integrity. Detecting current AI-generated texts and those written by the next generation of AI writers is going to require constant evolution, working closely with educators while training on available, relevant datasets.
We are reassured by the conversation around detection and other efforts to understand how AI writing impacts education, business, and every other aspect of society. There are risks to AI writing being used improperly, and it is essential that good detection technology is available across all sectors.
Turnitin is a global company dedicated to ensuring the integrity of education and meaningfully improving learning outcomes. For more than 20 years, the company has partnered with educational institutions to promote honesty, consistency, and fairness across all subject areas and assessment types. Turnitin products are used by educational institutions and certification and licensing programs to uphold the integrity and increase learning performance, and by students and professionals to do their best, original work.