Adapting GenAI for the next generation of learning
A new study by Monash University learning analytics researchers presents key considerations for generative AI (GenAI) educational tools so they are carefully developed to support, rather than replace, human learning.
Published in Nature Human Behaviour, the paper outlined essential factors for researchers, policymakers and technology companies to consider while adapting GenAI to support human learning in all levels of education and workplaces.
Key considerations included understanding how to use GenAI to enhance human learning while fostering skills for critical thinking and self-reflection in humans to effectively partner with GenAI.
The Centre for Learning Analytics (CoLAM) Director at Monash University’s Faculty of Information Technology and the senior author of the paper, Professor Dragan Gasevic, said powerful AI tools are set to become integral to society, transforming how we learn, work and live and GenAI technologies could permeate every aspect of human learning.
“Imagine students engaging in debates with digital twins of Socrates to explore ancient Greek philosophy, learning impressionist painting techniques from a humanoid robotic mentor modelled after Claude Monet, or visualising Einstein’s special theory of relativity in virtual realities,” Professor Gasevic said.
“This kind of integration needs a dual approach to learning: educating ourselves both about and with GenAI. This can be achieved through careful development of education tools informed by rigorous research and supported by unified efforts from education institutions, technologists and government policies.”
The study signalled that assessment processes should reward genuine knowledge and skill improvement over AI-generated illusions, that teachers needed support to adapt to the new GenAI landscape, and highlighted the need to promote human-AI interaction to maximise human learning while preventing over-reliance on GenAI.
The paper also emphasised that policymakers and tech companies must ensure accountability, develop appropriate ethical guidelines, and consider inclusivity when regulating and designing GenAI tools for education.
While AI tools can enhance learning processes and abilities they still present ethical dilemmas of transparency, privacy and equality, and have already caused disruptions in assessment processes.
The study's first author and CoLAM Research Fellow Dr Lixiang Yan said improving AI literacy for students and teachers alike is one of the crucial needs to be addressed to ensure the effective integration of AI into human learning.
“We anticipate a shift in educators’ roles, with GenAI reducing the burden of knowledge dissemination, allowing teachers to focus on deeper connections with students as mentors and facilitators,” Dr Yan said.
“Educational institutions must invest in ongoing professional development and support systems to help teachers manage techno-stress and workload burdens from adopting these new technologies.”
This research paper was a collaboration between learning analytics experts at Monash University’s CoLAM and researchers from the University of Luxembourg and Goethe-University Frankfurt.
The research was supported by the Australian Research Council, Australian Government through Digital Health CRC, Defense Advanced Research Project Agency, and Jacobs Foundation.
In addition to previous research on learning analytics, CoLAM experts are working on new projects to develop tools for assessing human-AI collaborative writing, improving knowledge sharing for educators, and enhancing workplace learning for healthcare professionals.
The researchers are also conducting a study with secondary students in 10 countries across four continents using pioneering GenAI tools to explain and enhance human skills in the age of AI.
CoLAM Director and co-author of the paper, Professor Dragan Gasevic, is available for interviews.
Journal Reference:
Lixiang Yan, Samuel Greiff, Ziwen Teuber, Dragan Gašević. Promises and challenges of generative artificial intelligence for human learning. Nature Human Behaviour, 2024; 8 (10): 1839 DOI: 10.1038/s41562-024-02004-5