ChatGPT: Should Lecturers Discontinue Take-Away CATs?

Published on 21st March 2023

The Challenge of Nurturing Thinkers and Innovators for the 21st Century

The President and CEO of Enactus, Robyn Fehrman, recently visited Kenya and met with university students and faculty advisors. Enactus offers university students the much-needed platform for social entrepreneurship and innovations. Her speech placed emphasis on deploying business and innovation skills among the youth as a promising way of impacting lives and livelihoods for shared prosperity in a sustainable world. This is the same world in which generative artificial intelligence is rapidly transforming ways of creating knowledge and developing skills. How will this affect the youth’s creativity and innovative capacity. On March 10, 2023, the Enactus Club of Taita Taveta University held an Open Day on the theme: Accelerating Sustainability and Entrepreneurship for Social Impact. The Annual Enactus Leadership Forum is also around the corner. This series of events places a lens over the new landscape of skills development and innovations as influenced by artificial intelligence. In the face of such rapid technological advances, we must ask ourselves: Will artificial intelligence be a catalyst or a catastrophe for nurturing the next generation of thinkers and innovators in the 21st century? This question has inspired an informed exploration of the unfolding challenges.

Key Highlights

  1. The COVID-19 pandemic was aptly characterised as a force of disruption, destruction, distraction, and division. In stark contrast, ChatGPT emerges as a ground-breaking innovation that fosters growth, sparks creativity, captivates minds, and bridges connections between people and ideas.
  2. With the entry of generative artificial intelligence, ChatGPT, college students now have access to a powerful tool that can assist them in researching and writing essays or solving take-away CATs. However, this has raised concerns among lecturers who see take-home assignments as a way of encouraging extensive reading, inquiry, and research among students. Questions have been raised about the future of critical thinking skills, grading fairness, and research ethics.
  3. To test the effectiveness of ChatGPT, which is based on OpenAI’s GPT-3 model and the precursor to GPT-4, a familiar question in Engineering Surveying was used. GPT-3 gave an incorrect answer, a result that pointed to the importance of human-level reasoning skills in similar technical subjects. GPT-4, however, was spot-on with the correct answer meeting the required accuracy level in engineering surveying, confirming the advanced reasoning capability that the GPT-4 model has proclaimed.
  4. While ChatGPT can serve as an accelerator to knowledge acquisition, it should not replace diligent immersion in the journey of learning. Thus, lecturers should avoid setting simple and straightforward questions that are easily answered by ChatGPT.
  5. Instead of viewing ChatGPT as a threat, we should embrace it as an opportunity to enhance learning outcomes through active, collaborative, contextualised, and self-directed learning. This is the spirit of Education 5.0, which embraces technology as a tool to democratise the learning environment. Thus, a lecturer’s decision to discontinue take-away CATs should be made after a balanced consideration of the resourcefulness of ChatGPT within the subject matter and the importance of critical thinking skills.

The Disruptive Entry of ChatGPT: Threat or Opportunity?

Generative artificial intelligence (GAI) is upending conventional wisdom on content creation and analysis. The grand entry in November 2022 of the generative pre-trained transformer-3 (GPT-3), known as ChatGPT, must have been sweet music to the ears of college students who hate spending long hours to research as the only way of writing quality essays or solving take-away continuous assessment tests (CATs). The subsequent GPT-4 model, just announced in March 2023, has even more breath-taking news for technology lovers and laggards alike. If COVID-19 deserved the description this thought-leadership and mentorship series gave it as a disruptive, destructive, distractive, and distancing disease, then justice can only be done by describing ChatGPT as a disruptive innovation that is creative, constructivecompelling, and connecting people and ideas.

The Disenchanting Questions

To lecturers who believe in take-home assignments as a way of encouraging extensive reading, inquiry, and research laced with flexible, self-driven discovery among students, ChatGPT has unearthed disturbing questions:

1. Will students take time to develop their own scientific inquiry skills if ChatGPT can do the job for them in minutes?

2. Will the grading of take-away continuous assessment tests (CATs) remain fair and accurately reflect the individual efforts of the students anymore?

3. Will we sustain a critical mass of young graduates who possess a genuine passion and philosophical persuasion for inquiry and critical thinking like Socrates and his students, as well as the thinkers who have followed in their footsteps in personifying logic and lucid argumentation?

4. How will intellectual honesty and research ethics be affected in a world where technology can easily facilitate fraud with elusive dexterity and subtlety?

The Democratising Lens of Liberation

To confirm if I should continue administering take-away CATs even after sounding a death knell for them when I interacted with Taita Taveta University students on the topic recently, I tested ChatGPT (GPT-3) using a familiar question in Engineering Surveying. The example was drawn from a tunnelling project. Here was the question:

For a straight section of the Northern Collector Tunnel in Kenya with a design slope of 0.155%, a 10-metre-long formwork/shutter was used. The upstream level of the formwork/shutter was at a reduced level of 1618.245 m. Calculate the correct staff reading at the downstream end of the formwork if a backsight reading of 1.025 m was taken to a TBM in the tunnel whose reduced level was 1620.155 m.

The correct answer to this question (awarding 9 Marks for all the steps and correct answer) is 2.951 m. ChatGPT gave 2.945 m in the first unguided attempt and then 2.926 m in the second attempt after guidance on how to determine the downstream level. Both answers are incorrect despite ChatGPT’s great attempt at using the established rules and formula for levelling exercises in surveying. The problem ChatGPT experienced was about reasoning, whether to add the level difference to, or subtract the same from, the known upstream level when determining the downstream level – and whether to use the TBM or the downstream level to calculate the staff reading. Download the verbatim ChatGPT and GPT-4 report from the link provided for more details:


When guided to note that the downstream level means a lower reduced level than the upstream level, ChatGPT gave an even worse-off answer, 617.615 m. It is GPT-4 that finally gave the correct answer without any hints as 2.9505 m (4 d.p.).

My moment of reckoning came, and it was both enlightening and liberating: students must understand that while ChatGPT can be a valuable tool for learning, it shouldn’t replace their commitment to actively engage in the educational journey through diligent immersion. Many technical questions will still catch lazy students off-guard and sleepwalking into intellectual poverty while enjoying some fleeting ChatGPT forays. This serves as a reminder to educators that “orchestrated immersion” is crucial for fostering a truly captivating learning experience. In today’s world, technology is indispensable for enhancing learning outcomes, promoting active collaboration, and fostering contextualised, self-directed learning. By incorporating multi-sensory, technology-driven methods such as story maps filled with contextual images, music, and videos, we can further enrich the educational experience and inspire innovations.

So was my aha moment. I can now state categorically thus: Depending on the way questions are set, lecturers of technical subjects like Engineering should still comfortably give take-away CATs. The trick is to avoid simple and straightforward questions that will be a cup of tea for ChatGPT. The integration of anti-AI writing detection capabilities into applications such as Turnitin is certainly promising news for educators who will still give take-away essays that require simpler problem-solving, less intricate calculations, and a reduced emphasis on logical reasoning and judgment.

Finally, rather than perceiving ChatGPT as a threat to effective learning, we should recognise it as a valuable opportunity – at the very least, a frenemy that can contribute positively to a flexible and democratised learning environment, epitomising the essence of Education 5.0. With the Knowledge Doubling Curve continually advancing, generative AI paves the way for accelerated knowledge acquisition, buttressing the importance of embracing the powerful technology tools unfolding in our eyes – a testament to the fact that competitive access to knowledge is the real power game of the 21st century. After all, making a difference hinges on upending outdated conventions and transcending the status quo, raising standards instead of merely adhering to existing benchmarks, and breaking records that have been made, including your own records.

By Nashon Adero

Lecturer and Enactus Faculty Advisor at Taita Taveta University
Email: [email protected]

This article has been read 1,519 times