
Last month we hosted an event discussing the impacts of generative AI tools such as ChatGPT on education. The session was chaired by The Data Lab’s Dr Alfred Tiley and we were joined online by teachers, lecturers, people working in the education sector, and Data Scientists/Engineers for an open discussion about challenges and issues currently being faced by educators.
There has recently been a growing concern among educators about using artificial intelligence tools in education. While some students may use AI to cheat on exams, others use them as assistive tools. However, those in attendance generally believed that if used in a constructive way, AI tools could have a positive impact on education.
The key points raised during the event have been summarised below (with a little help from ChatGPT!). At the end of this post, you’ll also find some valuable links and resources shared in the chat from those in attendance.
This discussion looked at the following:
- The need to update the education assessment process in a ChatGPT world
- The importance of the human in the loop
- AI and ChatGPT as a positive tool for education
- Adapting and preparing for a new normal with AI literacy
- The importance of teaching critical thinking
- The growing ubiquity of AI in education
- Institution policies and regulations on the use of AI
The need to update the education assessment process in a ChatGPT world
Attendees highlighted the need to align the assessment process with the current demands of the working world. They believed that students are using AI tools as a workaround and thought that the focus of discussions should be on how to address the issue of an outdated assessment system, rather than blaming the students for using AI tools, which are becoming the reality of the working world.
While some educators are concerned about the use of AI tools in education, others see them as a way to improve efficiency. Participants believed that institutions could adapt by incorporating AI tools like ChatGPT into their assessment processes.
For instance, students could use ChatGPT for coursework as long as they referenced it like any other source. Institutions could also look to detect plagiarism by comparing student output with their previous work and detecting sudden changes in language or quality.
One participant suggested that oral examinations may be a better way to test students’ understanding and critical thinking skills. Another participant suggested using tools like ZeroGPT to check whether a piece of work was written by an AI and adjusting assessments accordingly.
The speakers noted that several assessment models are effective and have no issues. However, some types of assessments need to be looked at more critically. These types of evaluations rely on written examinations to test knowledge retention, which is not always an adequate measure of learning.
Participants agreed that there needs to be a variety of assessments, including practical and continuous assessments, to gauge a student’s ability effectively. Overall, the speakers emphasised the need for a measured approach to AI and assessment design.
The importance of the human in the loop
Participants noted that the “human in the loop” aspect of teaching is essential in using AI tools effectively in education. Teachers need to understand their students’ capabilities and outputs to be able to detect sudden changes in language or writing style, indicating a potential use of AI.
The International Baccalaureate (IB) has recently announced that students can use ChatGPT in coursework as long as they reference it and include the prompt in the bibliography. However, the IB also stresses the importance of teachers being involved in the coursework process and building relationships with their students. By creating a close teacher-student relationship, teachers can see work take shape over a longer period. This allows teachers to detect sudden changes in output and more robustly detect cheating.
However, it was also noted that this approach may not always be feasible for specific instances, for example in university-level education where the class sizes are generally much larger and 1:1 educator-student time is minimal.
AI and ChatGPT as positives tools for education
Participants also noted how AI tools like ChatGPT could positively impact education, especially for students from areas of deprivation who cannot afford private tutors. For instance, ChatGPT could help students with research and writing, making their job quicker and easier. It could also make education more inclusive for students with disabilities.
Adapting and preparing for a new normal with AI literacy
Attendees agreed that it is essential to adjust to the new normal. Rather than rejecting them, educators need to embrace AI tools. AI literacy is vital for both students and staff. While AI has benefits, such as speeding up processes, it is necessary to understand the basics and go through the learning process to use these tools effectively.
AI literacy is needed to ensure that people know how to use AI and understand its limitations. It is important to continue teaching the fundamentals of reading, writing, and arithmetic, while also incorporating AI technology.
The importance of teaching critical thinking
Participants in the discussion stressed the importance of teaching critical thinking to students. The participants proposed strategies for teaching critical thinking skills, including having students engage in dialogues with chatbots and analysing the outputs generated by LLMs, such as breaking down structures, critiquing arguments and using AI-generated text as a learning opportunity.
The goal is to teach students how to work effectively with these tools and use them as a starting point for their work rather than relying solely on them. Participants noted that this was particularly important considering that the output from tools such as ChatGPT is regularly inaccurate.
Participants also mentioned that, in some cases, LLMs such as ChatGPT could be used to produce initial drafts of written work, such as essays, but emphasised the importance of students still thinking critically about the structure and content of their work. The participants believed that using LLMs to generate initial frameworks could save time and energy – as long as the correct usage of such tools was clearly communicated to students.
The growing ubiquity of AI in education
The discussion also covered the growing presence of AI in education, with examples of AI-powered tools being integrated into popular language learning apps such as Duolingo. The participants generally agreed that AI is here to stay, and institutions that ban using LLMs and chatbots are simply postponing the inevitable in learning to work alongside the technology.
The participants acknowledged that LLMs have limitations and can get things wrong, but they also believe these tools will continue to improve over time. They suggest that institutions should focus on teaching students how to work effectively with LLMs rather than trying to ban them altogether.
Institutional policies and regulations on the use of AI
The participants highlighted the need for policies and regulations to address the concerns about using AI tools in education. They mentioned that some institutions had developed policies around the use of AI, while others are still working on it.
They also discussed the need for regulation specific to the education sector, beyond general government regulation. Some participants suggested that policymakers should work closely with educators and students to ensure that policies are effective and well-received.
The participants also discussed specific examples of institutions and their policies on using AI tools in education. They mentioned that some institutions had taken a cautious approach, while others have adopted policies from other universities. For example, UCL recently released a policy on using AI in education that has been used as a model. Some institutions also look at how staff will interact with AI tools, not just students.
Interested in learning more about ChatGPT and generative AI tools? You might like our other articles:
Marking ChatGPTs homework with TDL’s Data Scientists
ChatGPT: Hype vs Reality – Answering our Event Follow-up Questions
Useful resources shared by those in attendance:
- GPTZero
- Acknowledging the use of generative artificial intelligence – Monash University
- Engaging with AI in your education and assessment – University College London
- Guidance for students on the use of Generative AI (such as ChatGPT) – University of Edinburgh
- AI & ChatGPT | StaffNet | The University of Aberdeen
- More Than Half Of College Students Believe Using ChatGPT To Complete Assignments Is Cheating – Forbes
- Half of College Students Say Using AI on Schoolwork Is Cheating or Plagiarism – BestColleges
Leave a Reply