AI Writing
5 AI Detectors Used by Colleges and Universities for Research
Several AI detectors are used by colleges and universities for research. Learn about the studies, researchers, and schools using them to study AI here.
While some universities and colleges use AI detectors to assess student work, others primarily use them for research purposes.
From exploring the effectiveness of AI content detection tools to determining AI use in academic publications, several AI detectors are used by colleges and universities to study the rise of artificial intelligence.
Read on to learn about some of the most popular AI detectors schools use for research.
Many AI detection studies coming out of colleges and universities use Originality.ai’s AI checker for research purposes.
As an exceptionally accurate AI content detector, it makes sense that Originality.ai is a popular choice among researchers who want meaningful results from their studies.
One notable example is an extensive study called RAID(RAID: A Shared Benchmark for Robust Evaluation of Machine-Generated Text).
In this study, researchers from the University of Pennsylvania, University College London, King’s College London, and Carnegie Mellon University evaluated 12 AI content detectors on shared benchmark data sets to determine their effectiveness.
The results? Originality.ai was found to be the most accurate AI detector on both base and adversarial datasets, as well as across all domains (types of content).
Also featured in the RAID study, GPTZero is another AI detection tool often used by university researchers. Originally built to distinguish between human and ChatGPT-written essays, it’s no surprise that those in academia want to put it to the test.
Take An Empirical Study of AI-Generated Text Detection Tools, for example. In this study, author Arslan Akram from The Superior University in Pakistan wanted to determine the effectiveness of AI detectors for multi-domain ChatGPT material.
In this study, GPTZero achieved a 63.77% accuracy rate on the Human and AI Text Database (AH&AITD). However, Originality.ai is the one that came out on top in this study with an accuracy rate of 97.09%, outperforming GPTZero by over 30%.
Not to be confused with GPTZero, ZeroGPT is another AI detector used by schools for research.
Claiming a 98% accurate rate thanks to its DeepAnalyse Technology, it has been a popular option for researchers who want to see if AI detectors are really that effective at distinguishing between human and AI-written text.
For example, ZeroGPT was one of the tools assessed in the study The Great Detectives (The Great Detectives: Humans vs. AI Detectors in Catching Large Language Model-Generated Medical Writing).
Here, researchers from The Hong Kong Polytechnic University and Rush University Medical Center wanted to see how AI content detection tools and human reviewers compared when detecting AI paraphrasing in articles.
ZeroGPT obtained accuracy rates of 96% (detecting ChatGPT-generated) and 88% (AI-rephrased) in the articles. It came in second to Originality.ai, which demonstrated accuracy rates of 100% (ChatGPT-generated) and 100% (AI-rephrased).
Turnitin’s plagiarism checker may be familiar to schools, but it also now offers an AI detection feature. So, naturally, it has also caught the attention of researchers studying AI writing, including those from colleges.
For example, William H. Walters from Manhattan College used Turnitin in his study on The Effectiveness of Software (The Effectiveness of Software Designed to Detect AI-Generated Writing: A Comparison of 16 AI Text Detectors). For his research, Walters compared the efficacy of 16 publicly available AI detectors in distinguishing between AI and human-written text.
The study found that Turnitin performed exceptionally well. The same study also highlighted that Originality.ai had 100% accuracy in identifying GPT-4 and GPT-3.5 text.
As well as Turnitin, the Copyleaks AI detector was also referenced in two studies:
In Recent Trend in Artificial Intelligence-Assisted Biomedical Publishing researchers from the University of Massachusetts Lowell, the University of the Cumberlands, and Miller Scientific, used AI content detection to identify trends in peer-reviewed biomedical literature.
Researchers highlighted that Originality.ai had a superior accuracy rate with an AUC of 97.6% (area under the receiver operating curve).
In this study, Copyleaks was noted as another type of AI detection software (along with GPTZero, Writer, and Crossplag) that could be applied in future research projects.
Although Originality.ai, GPTZero, ZeroGPT, Copyleaks, and Turnitin are some of the most common AI detectors used by colleges and universities for research, this is not a complete list.
The studies referenced in this article also looked at other AI detection software, including:
Read a meta-analysis of AI detection studies for further insight.
College and university researchers use several AI detectors in their research, whether they’re comparing their accuracy or trying to identify AI-generated content trends. However, when it comes to AI detection accuracy, one consistently comes out on top: Originality.ai.
With its ability to check for text generated by ChatGPT, GPT-4o, Gemini Pro, Claude 3.5, and more, Originality.ai is the go-to choice for those who need a reliable way of distinguishing between AI and human-written content.
Looking for more information about AI detection in education? Visit Originality.ai for educators to learn more!
