Brief Review – GPTZero

GPTZero Brief Review 

GPTZero (2024) , described as “bring[ing] transparency to humans navigating a world filled with AI content” on its homepage, provides educators an opportunity to “restore confidence and understanding” with students by detecting AI generated content. According to The Tech Advocate writer Matthew Lynch (2023), GPTZero is a significant resource due to its accuracy in detecting AI generated content to help maintain integrity in human-created works. Even though GPTZero is not not 100% reliable in detecting AI generated content, it still boasts a high degree of accuracy (Lynch, 2023). As I log in and initially explore GPTZero, I notice the AI scan is in the center of the dashboard; it is there, waiting to be populated with text. The text box also privileges the idea of finding human writing over sussing out AI writing by asking: “Was this text written by a human or AI?” The prompt also cautions users that if they are seeking to detect “AI involvement” – again, contributing to a narrative that human creation is privileged and not just identifying AI generated content. When compared to Turnitin’s (2024a) AI writing detection tool, which promises to “help educators identify when AI writing tools such as ChatGPT have been used in students’ submissions,” the goal seems to not be to penalize students it seems to be focused on first identifying what the student created on their own and opening up a conversation about when AI text has likely been used. GPTZero provides a “probability breakdown” of a spectrum of human and AI writing. Turnitin (2024b) provides a percentage that states “How much of the submission has been generated by AI” which is absolute. GPTZero emphasizes that it is not 100% accurate and leaves the implication open that a human needs to do more work to ascertain if there are any integrity violations. In fact, the report generated states specifically that the “report should not be used to punish students” (GPTZero, AI scan, January 28, 2024). One question I am wondering about the tool is: Where does the extensive database of human written content writing is compared against come from? 

References 

GPTZero. (2024). The best AI checker for teachers. GPTZero. https://gptzero.me/educators

Lynch, M. (n.d.). What is GPTZero? How to use it to detect AI-generated text. Tech Advocate. https://www.thetechedvocate.org/what-is-gptzero-how-to-use-it-to-detect-ai-generated-text/

Tian, E., & Cui, A. (2023). GPTZero: Towards detection of AI-generated text using zero-shot and supervised methods [Computer software]. https://gptzero.me/

Turnitin. (2024a). Turnitin’s AI detector capabilities. TurnitIn. https://www.turnitin.com/solutions/topics/ai-writing/ai-detector/

Turnitin. (2024b). Turnitin’s AI writing detection available now. Turnitin. https://www.turnitin.com/solutions/topics/ai-writing/

Week 14 Annotation 1 – OMMI: Offline multimodal instruction.

Dirkin, K., Hain, A., Tolin, M. & McBride, A. (2020). OMMI: Offline MultiModal Instruction. In E. Langran (Ed.), Proceedings of SITE Interactive 2020 Online Conference (pp. 24-28). Online: Association for the Advancement of Computing in Education (AACE).

            In this conference proposal, Dirkin et al. (2020) present a model of offline multi-modal instruction (OMMI) to address the digital divide made evidence during the acute phase of the COVID-19 pandemic. Dirkin et al. situate their instructional model in the context of datacasting, and provide an overview of the process. Data is securely transmitted one way by antenna to students so they can complete their learning. OMMI requires three separate elements to be successfully implemented: an anchor document, creative repurposing of software, and assessment that is performance-based. Anchor documents are students’ “home base” that give them the roadmap to success, and includes links to all elements students will need to study. Creative repurposing of software means using things like Word or PowerPoint to “mimic the interactivity of a website” by embedding videos and hyperlinking slides together. This provides students with an organized learning experience. Balanced assessment should be used because students may not have access to instant feedback from the instructor so formative and self-assessments need to be used to help facilitate learning. It is also important, given that students in the acute phase of the pandemic were under the care of parents who could not provide necessary levels of assistance to complete learning tasks, that students be given choices and scaffolded projects that were broken down in to small and workable phases. While Dirken et al. recognize it was best for learners to have access to the Internet, OMMI was the next best thing to help students not fall behind. The OMMI model replicates as best as possible the interactivity and interconnectivity of the Internet.

            Dirkin et al. organize their proposal very well. There is an immediate statement of the problem, followed by a contextualization in the literature and emergency of the COVID-19 remote learning environment. Dirken et al. ground their three-pronged approach in research and clearly define their operationalized terms to outline their argument. While conference proposals and proceedings are limited by nature, and cannot be fully developed due to space, the argument and support in the literature were very clearly articulated.

            As an administrator who had to navigate and survive the shift from a mostly on campus learning experience for learners to an all online experience, this work is interesting to me. While I do not work on a campus that largely did not have access to internet during learning, there were many challenges to the instantaneous pivot to remote learning that occurred in March 2020. For over a year, we had to find creative solutions to help learners where they were – particularly in hands-on disciplines like art and music, which I oversaw at the time. While this specific model OMMI is not directly applicable to my situation, the idea of the anchor document would have been extremely helpful in navigating learning in situations where a lot of learning was asynchronous. If I put on my writing instructor hat, I can see the benefit of preparing lessons and teaching opportunities, even as we have returned to “normal” – whatever that is – through an anchor document and creating interactive, non-Internet based resources to support learning for students who may not always be available to work online in the LMS.