As part of its #UOC2TheFuture and Sharing Learning initiatives, the UOC‘s eLearning Innovation Center(eLinC) organised a webinar with Mike Sharples, Emeritus Professor of Educational Technology at The Open University in the UK, to examine the effects of AI in higher education. UOC’s rector Àngels Fitó introduced the session which was moderated by Desirée Gómez Cardosa, from the eLinC’s Educational Trends and Innovation Observatory.
Since the beginning of 2023 artificial intelligence (AI) has featured prominently in the media. The news appearing in recent months has led to a feeling of unease regarding the constant new developments in the sector. To understand the changes that AI could mean for education, Emeritus Professor Mike Sharples has studied the uses of the ChatGPT large language model in academic contexts. The findings of his research directly concern teaching and learning in higher education.
AI in teaching and learning
Generative AI tools are both an opportunity and a challenge for higher education. According to Professor Sharples, large language models like ChatGPT can be used to produce exercises that focus on specific skills. One of the activities proposed by the professor is based on using AI to generate multiple responses to an open question. Students then critique the AI responses and create their own answers to the question.
Another exercise related to the teaching and learning process involves the research capability of AI. In this case, the educator sets a project for students to work in a group, using generative AI to research and solve problems. “We have to be very careful in critiquing research studies and making sure that they are valid,” he pointed out. Other uses suggested by the professor are related to personalizing the learning experience. Large language models such as ChatGPT can be used for Socratic dialogues, as guides, and for personalized tutoring. “It’s very important to prompt AI clearly and coherently,” he emphasized.
The role of the instructor
The appearance of AI tools has led to a discussion about the role of the instructor in teaching and learning. Professor Sharples stresses the importance of care in this new scenario. “Teaching is a caring profession, and yet AI is essentially careless and uncaring,” he pointed out. He feels that we should rethink written assessment, beware of AI and factual writing, as it’s not a database, and explore AI for creativity, argumentation, and research to help students develop the skills of critical thinking.
Although new AI systems are constantly being developed, Professor Sharples recommends universities adopt models based on ethical criteria, such as Claude. This generative AI tool, developed by Anthropic, has been developed from the outset on ethical principles. “I think there’s a real opportunity to put pressure on the big companies to develop their models on similar ethical principles and for universities to make a positive choice to choose ethical AI,” he continued. As he pointed out, these models were aimed at the general public. “Generative AI based on ethical principles is more financially accessible and willing to indicate the data it uses,” he added. In the professor’s opinion, there is an urgent need to develop a new AI literacy so that the use of these systems will benefit students in higher education.
In the last part of the webinar, questions were invited, and Professor Sharples gave his opinion on the following topics:
- The hallucinations of AI as a mechanism for creating innovative ideas
If, instead of seeing generative AI as a database, you see it as a creativity engine, then the hallucinations can, to some extent, be valuable. There’s also a setting on ChatGPT and other systems called temperature, and it’s like a dial that you turn. The higher the setting, the more unpredictable it becomes. If you want it to write a technical report, you might dial down the temperature so that it becomes more predictable and more conventional.
- Regulating AI globally
The professor highlighted two issues: what sort of constraints should be imposed on the use of inappropriate language and how you build them from the bottom up on ethical principles. Language models can be built on principles of individual empowerment, such as the United Nations Declaration of Human Rights, to choose ethics over irresponsibility. There is an opportunity for universities to adopt ethical AI systems that are more open in terms of indicating what data they’ve been trained on and more open in terms of allowing students wider access and developing AI literacy.
- Will AI boost adaptive learning and personalization in the learning process?
Professor Sharples thinks we need to be very careful about personalized tutoring systems using generative AI because of the danger of them making mistakes, and this is all part of AI literacy. He said that there will be a new breed of what is called hybrid AI systems that combine symbolic AI, reasoning systems, and databases with generative AI.
- In the long term, what is the most significant threat from AI in the classroom?
Professor Sharples believes that the biggest danger in education is that it will devalue human teaching if the teacher is seen only as a transmitter of knowledge and information. It’s up to us to treat AI with care, not trust it, and also to recognize the new role of a teacher in terms of providing a much more personalized and careful interaction with students. There are many ways in which the human teacher is absolutely essential to the learning process.
- Teachers and their ICT and AI skills
There’s a huge need and urgency for developing AI literacy. AI literacy is about first understanding how AI systems work. The first thing is for teachers to have a basic understanding of how AI systems work, and secondly, to get their hands dirty and, for instance, actually play with them to develop practical competencies. If they do this, they won’t be afraid to use them as tools for their own workflow, for instance, for creating lesson plans or material that students could use. Universities can work together to develop a set of principles for AI literacy.
Last thoughts and conclusions
The future of higher education will largely depend on how the different agents involved deal with the emergence of cutting-edge technologies. Their incorporation in the classroom and online is fundamental, as are teaching methods and course design, but we should not disregard the importance of education policies and ethical and legal frameworks. There are many obstacles to be overcome in a society where we still have to tackle challenges such as the digital divide, train teachers in the latest ICT, and engage other relevant parties.
Yet another challenge is posed by generative AI systems, which are revolutionizing education by undertaking tasks normally performed by teachers and students: writing assignments, summarizing texts, planning courses, and engaging in dialogue. The appearance of different types of AI has prompted an urgent need to regulate their use in classrooms, with special attention to assessment and plagiarism.
Professor Sharples has told us about the capabilities and limitations of current generative AI. From the information presented in his webinar, we can see clearly that the role of educational institutions must be inclusive, and the challenge facing them is to assimilate today’s social and technological change for the benefit of society as a whole.
To conclude, let us highlight one of the professor’s main ideas: rather than seeing AI as a threat to traditional education, we should focus on preparing students for a future in which it is a tool for creativity, which we will need to use carefully, aware of its limitations.