Can GenAI generate a new pedagogy?

With all its abilities, a big question is if Generative AI (GenAI) can reshape the way education has been imparted over the centuries? The short answer is yes, but not immediately. The promise is there for everyone to see, but there are still plenty of kinks to iron out and guardrails to install for the technology to be truly effective in learning outcomes.

Educators, education service companies and, of course, students have started using GenAI tools for their advantages such as scale, ease, efficiency and adaptability. Some of the apps and tools such as ChatGPT (which is an LLM-based chatbot that can help in creating content for various uses), TutorAI (help students learn about any topic or subject) and Poe app (a mobile fast app for themed AI study bots) stimulate creativity by assisting students in brainstorming and generating ideas, thereby providing a fresh perspective and help students explore innovative concepts. Additionally, given that the answers depend on the questions asked, these apps and tools also help students learn one of the most important skills - how to ask better questions. 

In the US, teachers have started using GenAI models as assistants to create application-based homework and assignments, explain simpler concepts to their students while they focus on the more complex portions of the lessons as well as for their own academic research. Initiatives such as Dave AI provide teachers in K-12 education with AI-driven activities that can be incorporated across different subjects. Additionally, universities like the University of South Florida are investigating GenAI tools to enhance student learning. These technologies assist in tasks like brainstorming, offering writing feedback, and tailoring learning experiences to individual students. Even Khan Academy has created a special tutor called Khanmigo, which helps students think and challenges them but does not give them the answers straightaway. 

Apart from helping teachers and students in improving productivity and managing time more effectively, these tools also allow teachers to personalize education Duolingo, for example, uses GenAI to create personalized learning plans for each student thus helping create an engaging learning experience.

At a more macro level, even countries have started using GenAI to completely rethink their education systems – from curricula and teacher training to classroom lectures and even examinations. Several countries (such as China and Finland) are slowly and carefully using pilot projects to see how they can harness technology for maximum benefit. While China is using tax incentives to encourage the adoption of tools like the adaptive tutoring platform Squirrel AI, in Finland, approximately 50% of schools have adopted the ViLLE platform, which offers instant feedback and analytical insights on student assignments to both students and teachers.

Despite its potential promises, GenAI also has its perils. Generated content may vary widely in coherence, accuracy and overall reliability, making it challenging to rely on for critical tasks. The tendency to sometimes make up answers confidently even when they are factually incorrect and to hallucinate and see patterns where none exist are well known. In addition, there are issues such as model bias (at data or algorithm levels); copyright issues with data being used to learn from; as well as privacy issues because the model is collecting users’ questions and data, learning their insights, and thinking process. 

Additionally, LLMs are not cheap – at least not the ones that will give you correct answers most of the time. It is a resource-intensive process that requires high-end GPUs to process large volumes of data. The process usually spans many months and needs proper infrastructure, highly skilled engineers, and investing in the right algorithms, to name a few requirements. While many corporations are able to absorb the costs, it is yet to be seen how well the educational sector can deal with it.

The biggest worry though is of access. Given the digital divide, will adoption of GenAI only increase the gap between the digital haves and have nots?

The advent of GenAI in education heralds a promising yet complex future. While it offers vast potential to transform the methods of teaching, learning and assessing, there exist substantial hurdles still. As educators and policymakers explore the possibilities, it is crucial to tread cautiously, addressing these challenges while harnessing the GenAI’s transformative power to create a more inclusive, effective and equitable education system for all. Only through thoughtful implementation and continuous refinement can we unlock the full potential of this technology while mitigating its inherent risks.

The article was first published in The Financial Express on 13 May 2024.