Can artificial intelligence (AI) be used to support human teaching and assessment practices? This is what Dr Qian Xiao asks in her HUMAN+ research project. Using machine learning techniques to generate information for new forms of teaching, Xiao explores the potential of intelligent tutoring systems, or ITSs: computer systems that provide direct feedback to students to complement human teachers’ judgement. Xiao is part of the HUMAN+ programme, led by ADAPT, the Science Foundation Ireland Centre for Digital Content Innovation and the Trinity Long Room Hub Arts and Humanities Research Institute, to promote human centric approaches to technology research.
Focusing on student performance, Xiao explains, would be the first step in establishing an education system that can assist instructors in assessing students. Assessment could be calculated automatically by an algorithm, giving educators time and space to focus on individual students’ habits and behaviour. Often even strong students have differing learning requirements, and this research would enable teachers to direct their attention towards meeting these needs.
The HUMAN+ programme was the ideal destination for Xiao’s research at the intersection of education and AI when she arrived Ireland in 2019. When searching for positions that would enable her to work in AI from Dublin, TCD came up as the top result. She then contacted Prof. Vincent Wade, one of the two project co-ordinator managers of HUMAN+ and the chair of Artificial Intelligence in the School of Computer Science & Statistics, to discuss her proposal. Prof. Wade would soon become her Human+ science supervisor, alongside Prof. Keith Johnson, from the School of Education, as her humanities supervisor.
In comparison to her HUMAN+ colleagues, Xiao’s current research is a significant departure from her PhD in data privacy, which also led to a role as a Senior Engineer in Huawei Technologies in China. This meant the first few months in her fellowship were spent exploring an entirely new area. Xiao welcomed this challenge, as a pervading theme throughout her career has been problem-solving: she considers it more important to be able to adapt to changes in the fast-moving field of Computer Science than have a background in a specific area, and HUMAN+ provided her with the tools to discover a new area of research.
Xiao’s work builds upon an app released by a Korean company which recorded the student exercise history of almost 800,000 users for two years. When she and her supervisory team discovered the release of this dataset, they decided to see if it could be used in machine learning to create an ITS. Discussing this research, she compares students’ output to predicting the stock market: “think about each student as a company.” While this was a useful starting point for Xiao, it was also a heavily concentrated area of Computer Science that had been widely examined with very sophisticated algorithms for the last six years. As a result, she decided to redirect her focus: instead of predicting a student’s next answer, she elected to examine students’ long-term performances. This move is what makes her work stand out in terms of innovation, according to science supervisor Prof. Wade, who informed us that
So, does Xiao’s work line up with the HUMAN+ ethos of focusing on the human element of research? “Often, when I am asked about ITSs, people think ‘robots,’” she says. “But if you think about your favourite teachers, they aren’t just good at assessing. Their biggest advantage is they are good at encouraging and supporting their students. What we are doing is setting the teacher free, so they don’t need to spend time on assessments.” From a human-centric point of view, then, Xiao wants ITSs to provide a novel avenue for human instructors, presenting a qualitative way to remove bias and offer a new perspective on assessment.
Working with two supervisors from two different fields, Xiao’s fellowship is creating new connections between Computer Science and Education. After spending 4 or 5 months with Prof. Wade exploring the data, Xiao approached Prof. Johnson who provided feedback on the viability and innovation that it would offer to educators. This type of inter-departmental communication was not something Xiao had encountered in a structured programme before, yet it has been key to the success of her research and highlights the importance of its multi-pronged approach to human-focused technological research.
Indeed, discovering how to present findings to a wider audience has been a novel focus of Xiao’s first year with HUMAN+. This has included attending TCD’s Tangent courses on innovation pathways for researchers. Tangent courses teach innovation pathways, communication skills, and leadership, which have provided Xiao with the tools needed to present Computer Science findings in a way that is accessible for non-experts. In her HUMAN+ tech talks, she is exploring how to communicate her research without the field-specific terminology she is used to. As she notes, explaining the specifics and requirements of an ITS requires a lot of imagination and understanding when it comes to fields where algorithms are rarely given any attention.
Xiao believes this work is paying off, as her team is now in a position to approach industry partners. “Right now, we have something to show we have strengths and expertise in this area, so we can broach the idea of collaboration.” Xiao sees her current project as part of a longer-term vision with significant ramifications in the world of education. Any industry connections are unlikely to be only short-term, as there will be significant long-term benefits that will span beyond the timeline of her own fellowship.
Reflecting on her experience as an international researcher, Xiao highlights what she sees as the differences between research in Ireland and elsewhere. Asian universities focus on benchmarks and excellence in algorithm performance, while US institutions are constantly looking for new problems to solve, often via collaboration with industry. In Ireland, in comparison, because of the ubiquity of research funding from governmental and EU sectors, there is a larger focus on mental health and human-centred research. This highlights Ireland’s research interest in what Xiao calls “tangible benefits to real life, rather than just benchmarks or commercial problems.”
It’s no surprise, then, that Xiao’s fellowship is taking place in an Irish university and indeed an Irish programme where human-centred research is valued. Her work ultimately aims to provide educators with more time for interaction with and support for their students. As she herself observes, “Just confirming if you can answer a question correctly will not give any tangible benefits. We need to discuss the human-centric benefits. And not just about the technology. After we have the technology we need to see the changes it can make to a human’s life. I think that’s what makes the difference.”
Advertisement: PhD Studentship in Harmonised Semantic Model of EU AI and Data Protection Regulations. Supervised by @coolharsh55 and @EdoCeles at @AdaptCentre in @DCU. Details and application: https://www.adaptcentre.ie/careers/phd-studentship-in-harmonised-semantic-model-of-eu-ai-and-data-protection-regulations/ Please repost 🙏