AI can produce increasingly human-like writing, pictures, music and video. There have been reports of students using it for cheating, and an industry has emerged around AI-authored books claimed by people as their work. Students using AI tools such as ChatGPT to complete their work is one of the major concerns in the academic sector.  

One of the main advantages of artificial intelligence language models is that they provide a platform for asynchronous communication. This feature has been found to increase student engagement and collaboration, as it allows students to post questions and discuss topics without being present. 

 Assessment is an integral part of higher education, serving as a means of evaluating student learning and progress. There are many different forms of assessment, including exams, papers, projects, and presentations, and they can be used to assess a wide range of learning outcomes, such as knowledge, skills, and attitudes.   

While ChatGPT has the potential to offer many benefits for assessment in higher education, there are a few key challenges that ChatGPT, and other artificial intelligence language models like it, may pose for assessment in higher education. One challenge with using GPT-3 for assessment in higher education is the possibility of plagiarism. AI essay-writing systems are designed to generate essays based on parameters or prompts. Students could use these systems to cheat on their assignments by submitting essays that are not their work. 

AI detecting AI 

 Students trying to cheat is not something new. But the presence of AI tools has strengthened the gravity of the issues. While there are AI tools that help us create content, other AI tools detect if the content is human-generated or AI-generated. 

 Teachers and educators already have existing tools and strategies that could help them check essays. Softwares such as Turnitin, Copyleaks, Sapling, Winston AI etc., are such tools. The ChatGPT creators OpenAI has also contributed to some counter tools such as GPTZero and AI classifiers to the market. 

 An AI can detect another AI using pattern recognition. The checkers use unique identifiers that differentiate human writing from computer-generated text. "Perplexity" and "Burstiness" are two metrics in AI text-sleuthing. 

Perplexity measures how well a language model performs in writing good, grammatically correct, possible sentences- in short, how well it predicts the next word; Burstiness refers to the variance of the sentences.  

However, AI is getting even better at sounding human. In a recent paper by Stanford reseats GPT detectors showed bias against non-native English writers.

Tip for university staff 

University staff can use a few key strategies to design assessments that prevent or minimise the use of ChatGPT by students. One approach is to create assessments that require students to demonstrate their critical thinking, problem-solving, and communication skills.  

Academic writing is expected to accurately cite and reference the work of others, including in-text citations and a list of references at the end of the document. This helps to give credit to the original authors and supports the research's validity and reliability. Output from ChatGPT or other AI language models may not include proper referencing, as they may not have access to the same sources of information or be programmed to correctly format citations and references. 

Another strategy is to create open-ended assessments and encourage originality and creativity.  

Conclusion 

 Current AI tools have the potential to offer a range of benefits for higher education, including increased student engagement, collaboration, and accessibility. ChatAPIs can facilitate asynchronous communication, provide timely feedback, enable student groups, and support remote learning. At the same time, GPT-3 can be used for language translation, summarisation, question answering, text generation, and personalized assessments, among other applications. However, these tools also raise numerous challenges and concerns, particularly academic honesty and plagiarism.  

 Universities must carefully consider the potential risks and rewards of using these tools and take steps to ensure that they are used ethically and responsibly. This may involve developing policies and procedures for their use, providing training and support for students and faculty, and using various methods to detect and prevent academic dishonesty.