Dear all,
Generative AI, particularly Large Language Models (LLMs), is undoubtedly disrupting higher education and challenging traditional assessment formats and methods due to risks of cheating, ethical and data security implications, and/or a lack of educational value (e.g., lack of assessment for learning impact). Does anyone have knowledge of and can recommend empirical research with concrete methods, formats, and/or recommendations for addressing this challenge? For example, studies that have actually validated and/or tested the effectiveness and/or usefulness of these methods, formats, and recommendations.
Many thanks in advance.
Best,
Mikkel
It is pertinent question Mikkel Godsk. Currently, there has been a lot of discoursive work and essay-type papers about Gen AI. However, there is a dearth of empirical papers since we are still trying to figure out ways of grounding research with a completely new approach to generative AI.
One of my papers is in review in a management journal where we have developed a method to use different Gen AI technologies in text analytics applications. Would be happy to share the paper once it gets accepted.
Yes, I completely agree and I am very much looking forward to seeing this type of research available Sharif Ahamed.
Your paper sounds interesting Arindra Nath Mishra. Will it also address the challenge of assessment?
I am afraid, my paper may not address it directly. However, I have been using a tool called ZeroGPT for assessing AI-based plagiarism. There are some other tools as well like Turnitin that give a score for AI-generated text. Over time due to regulatory pressures, Gen-AI would include some fingerprints that could be more conclusively detected.
In the long run, we need rules for how and when to use Gen AI in an academic context. The biggest problem is that not enough focus has gone into the dark side of this technology that you have mentioned. Issues such as cheating, privacy violations, and degradation of educational value need to be addressed parallel to the development.