Despite it be the online environment or not, the assessments should always be aligned with the intended learning objectives and planned accordingly to assess higher order cognitive skills. The traditional view point of testing knowledge has shifted to assessing skills and application. In order to achieve this there are multitude of assessment techniques implemented.
In the current blended learning environment platforms are improving and assessments are increasingly becoming online. These might be of a broad spectrum from traditional MCQs and essays to blogs and projects.
LLMs (large language models) or AI guided writing tools could cause a huge impact in student assessment modalities. The field of LLMs demonstrated an accelerated development since 2017. Launch of Chat-GPT last year was a landmark as it showcased immense ability to generate content equivalent or surpassing human abilities to user inputs. Despite certain limitations it showcased the ability to conduct certain tasks with out need for human intervention.
The potential of it to threaten the usual way of assessment was demonstrated by Chat-GPT passing medical and law entrance examinations. In the practical world what is most worrisome is the impact these platforms can cause on assessments where students creative and writing skills are tested ,rather than in ‘standard exam’ type platforms where access to these programs can be blocked.
In written assignments , projects and blogs etc . students might take an easy pathway to use LLMs. LLM work may not be detected by plagiarism checks .Additionally certain versions of LLMs are accessible through paid subscriptions. These platforms do have better performance and may pose a differential advantage to students who could afford to purchase them.
In addition to aligning the assignments with learning outcomes , this demands broader consideration ,where the impact of LLMs on assignments should be taken into account when designing them . Current day educators will have to face the challenge of finding was to assess students ‘bypassing’ the differential advantage of using AI writing tools.
Certain strategies would be to format the assignments in such a way where application skills in a more practical aspect is tested , and allow students to use AI writing tools in their work and to indicate how and when it was used.
In summary , as modern day educators ,we are unable to disregard or ban LLM usage , rather should focus on incorporating them in our assessments in efficient manner.
May 12, 2023 at 11:40 am
I found your blog post to be very thought-provoking. I think that you correctly highlighted the shift from traditional knowledge testing to assessing skills and applications, noting the diverse techniques used in today’s blended learning environments. Your discussion about the potential impact of Large Language Models (LLMs) on student assessments was particularly compelling. The rise of LLMs like Chat-GPT indeed presents new challenges for educators in preserving the integrity of assessments, particularly written assignments. It’s clear that these AI tools may offer an unfair advantage to students who can afford them and their use is hard to detect with standard plagiarism checks. I also found your suggestion to embrace LLMs rather than trying to ban them compelling. Designing assignments to assess application skills in practical aspects, and allowing students to declare their use of AI tools could be a viable approach. Your blog poses important questions about the evolving landscape of online assessments in the era of AI!
May 15, 2023 at 8:18 pm
Thanks Natasha for this thrilling blog post. It is worrying to hear that Chat-GPT passing medical and law entrance examinations so it can jeopardise the current assessment methodology. At this point, I agree with your attitude and try to utilize Chat-GPT like technology to update our assessment methodology and utilize it without disturbing fairness in the evaluation.
May 25, 2023 at 12:37 pm
Hi! Your reflection on the challenges posed by AI related tools like ChatGPT is definitely topical and the urgency of this question also makes many of us wonder whether we should rethink our methods of assessment as well as design of course assignments. What different and alternative models of assessment as well as assignments can one design that move beyond the traditional forms and modalities for learning and examination? Something we will continue pondering…/ Gül