AI can enhance higher education - but must not replace critical thinking - IOL
As AI becomes part of everyday learning, educators face a growing challenge: how to use technology without weakening critical thinking.
ARTIFICIAL Intelligence (AI) in education should not compromise teaching and learning; rather, it should enhance it.
It is important for both students and academics to empower themselves with AI to get the best out of it. For example, the vision of the University of KwaZulu-Natal is to become an “African Scholar,” and its mission is “to engage critically with society”.
This is impossible if we let AI take over our reasoning and judgement. To be a scholar, one is still expected to be knowledgeable. To critically engage with society, one cannot outsource the reasoning capacity to AI.
I attended an international virtual session organised by the UKZN where one presenter said, “AI will not replace you; rather, people who are more knowledgeable than you on AI will replace you”.
I sometimes see how, with the limited knowledge I have of AI, I’m able to use AI to produce, for example, a training schedule for short-term executive courses that wows my colleagues because they don’t yet know how to use AI in that area.
How you empower yourself with AI is to gather as much information as you have access to by reading about AI, AI articles on the internet, on social media, in online newspapers, and attending workshops or webinars on AI. The more you get to know how to phrase your prompts in such a way that you are as specific as possible, the more you get exactly what you want from AI.
You are also able to guide AI by giving it a sample of what you want by attaching it and asking AI to reproduce it in the exact format. For example, you may ask AI to review a research proposal using the attached template.
UKZN is one of the universities in South Africa that has developed guidelines on the use of AI in academia. They allow students and academics to use AI ethically and with integrity. For instance, the guidelines require users to verify what is generated by AI with valid reference sources. This is not a new requirement in academic writing.
Before AI, in academic writing, one was required to always acknowledge the source of information by referencing. That requirement continues during the era of AI.
This requirement becomes more important because the information AI generates may not state or provide reliable sources for the information it produces. It therefore becomes the responsibility of the person using AI to generate information to find those sources and acknowledge them by referencing them.
The other reason why it is important to verify the information generated by AI is that AI sometimes hallucinates. That means AI generates incorrect information.
AI systems can sometimes express uncertainty or indicate that they do not know, but they can also produce incorrect information confidently. So, how will one know if the information generated by AI is incorrect without verifying it?
The other way of getting the best from AI is feeding AI with information rather than asking AI to generate the information for you. Maybe ask AI to refine and improve the information you gave it. In that way, you are able to pick up on when there are mistakes.
Another effective way of using AI is asking it to simplify information for better understanding. Because you know the information, you may pick up on the simplification by AI if there are any errors.
For instance, I asked AI to simplify the ABC control system combined with the Pareto analysis in inventory management and provide an example.
It provided a simple explanation and gave me an example of a bookshop. But there were mistakes with the allocation of percentages in the Pareto analysis. Because I knew what the content was all about, I was able to pick up the error.
Yet, if a student did not bother to read about the ABC control system combined with the Pareto analysis in inventory management or listen attentively during class and went straight to AI to generate information, the student wouldn’t be able to pick up the error.
This has implications beyond how we academics use AI ourselves; it also requires us to rethink how we assess students.
Assessments have to evolve and be designed, as far as possible, to be resistant to AI-generated responses. While students are permitted to use AI, we cannot allow them to outsource their thinking to it, as this would undermine the very purpose of higher education, which is to produce critical thinkers.
One way of achieving this is through authentic, real-world assessment: students could, for instance, be asked to investigate an existing company and analyse the sustainability of its supply chain.
Because the company and its supply chain can be independently verified, this kind of task makes it considerably harder for a student to rely on AI to fabricate a fictitious company with an equally fictitious supply chain.
Oral assessment is, for similar reasons, gaining renewed value in this era of AI.
Unlike a written submission, an oral assessment allows the examiner to probe a student’s understanding in real time, distinguishing genuine comprehension from AI-generated competence. Class participation is, as a result, carrying more weight than it once did.
There is even growing discussion, in some quarters, of introducing an oral defence of the doctoral thesis in place of, or alongside, the traditional written submission.
This is, admittedly, not an entirely new idea; oral defences already form part of doctoral examination processes at a number of universities in South Africa and abroad.
* The views expressed do not necessarily reflect the views of IOL or post.co.za
