The introduction of generative artificial intelligence tools such as OpenAI’s ChatGPT and Microsoft’s Copilot is forcing South African schools and universities to rethink the way they assess pupils and students.

Assessment criteria that rely on information recall – and even more complex cognitive tasks like essay writing – are no longer effective at gauging comprehension since students can use a simple prompt to get a compelling answer from AI.

“A number of assessments are being redesigned,” Diane Grayson, senior director of academic affairs at Wits University, told TechCentral.

“For example, assessments that just require students to write an essay are having to be modified to prevent them from bypassing the intended learning experience. Instead, more assessments are being given that require students to use AI tools and to show how they have engaged with them, such as critiquing AI-generated text, or showing how they used an AI tool to assist them with an assignment, including giving the prompts they inputted into the AI tool.”

The incorporation by educational institutions of AI into their assessment tasks is, in part, a form of cooperating with the inevitable: students will use the tools at their disposal, whether teachers want them to or not.

But AI cannot – and arguably should not – be used in every assessment. In many instances, a gauge of the student’s raw capacity is more important to educators than knowing how well they can use a particular tool – think manual arithmetic versus calculators, for example.

Real world

In face-to-face learning institutions, it’s easier to prevent the use of AI tools through monitoring to ensure students comply with such directives. This, however, is rarely possible in distance-learning institutions, which rely heavily on online testing and written submissions for carrying out their assessments.

Unisa has already incorporated AI into its exam proctoring software suite to combat AI cheating.

“Unisa continues to take proactive measures and continues to use proctoring and AI tools such as the Invigilator app, Moodle and Iris to secure academic integrity. We also promote the usage of Turnitin, web-based text matching software that has an AI content detector as one of its measures to protect academic integrity,” Unisa said in a statement on Friday.

Another reason why higher learning institutions are driven to incorporate AI into their teaching and assessment practices is that their students – who are soon going to be employees – are better prepared for the real world if they are taught how to use AI to assist them in their work.

However, a similar argument applies to pre-tertiary learners, too. School teachers are just as concerned about AI cheating as their counterparts in tertiary institutions.

Independent Examinations Board (IEB, or private) schools had concerns of academic dishonesty that could result from internal assessments, said Confidence Dikgole, CEO of the IEB. “The IEB responded by convening a think-tank consisting of teachers and heads of schools, where they raised concerns about AI. The IEB utilised an industry expert from the tertiary sector who gave insight into how universities were grappling with the same challenges. In response, the IEB has created an AI usage guideline.”

Up-to-date policies and guidelines help teachers navigate the challenges new tools bring to the learning environment, but if teachers are to apply them judiciously, they must have a working understanding of the technology themselves.

According to Dikgole, teachers use assessment feedback to determine how learners are progressing in their learning – if this feedback is not valid, teachers may make inaccurate conclusions on their pupils’ progress. She said the assessment literacy of teachers is becoming increasingly important and teachers are required to develop assessments that incorporate higher-order questions and transversal skills.

Tech savvy

“Teachers must remain tech savvy so that they are aware of both the challenges as well as advantages of generative AI in their teaching and assessment practices,” said Dikgole.

She said AI brings with it opportunities for personalised learning that could help the system of “mass education” reimagine how teaching can be delivered in ways more appropriate to individual pupils. Adaptive assessment and automated marking are also among the new possibilities brough about by AI tools.

Wits’s Diane Grayson cautioned, however, that schools should ensure that AI tools are deployed equitably so that they are available to all students, despite their financial means or special needs. “The more powerful AI tools are not free, which exacerbates the digital divide between students with different financial means,” she said.

Grayson also cautioned teachers and institutions to use student feedback as a tool to improve their approach to using AI in their teaching and assessment practices.

“Both students and lecturers are in the early stages of learning about what AI tools can and cannot do and how best to use them to teach more effectively and support rather than circumvent or even undermine learning. Embracing AI tools in education is a journey that everyone in education has recently embarked on, and the progress on the journey will need to be constantly monitored and practices will need to be constantly adapted as everyone learns and grows along the way,” she said.  – © 2024 NewsCentral Media