Artificial intelligence feeds natural brains
Leibniz researcher Daniele Di Mitri has been elected as AI Newcomer 2021.
17.02.2021 · Humanities and Educational Research · DIPF · News · HP-Topnews · People
How can Artificial Intelligence be used in education? Daniele Di Mitri is working on this at the DIPF I Leibniz Institute for Research and Information in Education. (Update: In April, he has been elected as AI Newcomer 2021, called by the German Informatics Society.) An interview about distance learning, digital teaching beyond video conferences and why we should not only use Artificial Intelligence to display suitable films or advertisements.
Interview: Christoph Herbort-von Loeper
Leibniz: Digitisation in education is a hotly debated topic, especially in the pandemic. But the discussion is more about overcoming spatial distance and your field of research, Artificial Intelligence (AI), is rarely discussed - why?
Daniele Di Mitri: The debate is very much focused on overcoming physical distance. Distance learning is often limited to fairly simple forms of digital teaching, where students follow a video conference as mostly passive viewers. Before we talk about AI in educational practice, the potentials of digital education should be better exploited in distance or blended learning, such as with collaborative online formats, the use of e-learning platforms or asynchronous discussions via online forums.
Where do you see realistic application scenarios between elementary school and professional development?
Ironically, AI is only widely used in educational contexts to track down plagiarism, and there are so many more productive ways. AI can help teachers to find relevant sources for their lessons, chatbots can be used for both for question answering and for retrieving information, as well as support for the well-being of learners in general. Automatic summaries of content, the conversion of text into learning podcasts that can be listened to in a flexible manner, or basic corrections of essays on questions of spelling and grammar are also conceivable. Looking a little further into the future, I can imagine intelligent tutor systems that adapt to the level of learning and thus provide individually adapted tasks and feedback, for example in math or statistics.
AI in school - that is likely to create horror in many parents. Can you reassure them?
I can understand such fears related to new technologies. Even with the introduction of television, many people had feared the end of education. It is, however, the case that we already use many AI-based applications, for example when we get suggested content from Netflix or Spotify, or use Siri, Alexa and Co. Of course there are negative examples of data misuse - new technologies always have positive and negative effects and data protection must be an important factor, but on the other hand: Why don't we use the possibilities of AI for something more meaningful than advertising? The pedagogical concepts that we support with AI should be decisive for our assessment.
Many schools in Germany already reach their (technical) limits with distance learning. How realistic is it in practical implementation?
Schools - teachers and students - were confronted with a huge challenge by the corona pandemic, which has thrown the previous forms of teaching almost all over the place. I don't blame the teachers at all for this when it comes to the state of digital education. They are really doing everything they can to deal with the situation. We need - completely independent of AI - a comprehensive concept for a digital learning infrastructure to tackle digital divide not only in Germany: functioning learning management systems, secure cloud storage available for teachers and learners, e-mail addresses for all students for secure communication, a waiver of non-verified apps to protect their data and significantly more IT Experts in schools. In the economy, roughly one IT expert is expected for every 30 employees; for schools that would mean one IT expert per class.
Where do you want to be with your research in ten years?
The development of AI is proceeding so quickly that in ten years we will probably have possibilities that we can hardly imagine today. Above all, I expect that hybrid systems, where humans and machines collaborate, have become much more widespread. My hope is that we have collected empirical data to advise governments and policymakers on where AI can be meaningfully used in education. Multi-modal tutors who can also provide assistance in practical learning, not just theoretical skills, by the use of Virtual Reality. However, there are still many open research questions, both on the technical and pedagogical side, e.g. how should the AI’s feedback on the learners look like, should it, for example, allow mistakes from which one can learn or interrupt beforehand? I would like to contribute scientifically to these aspects.
Daniele Di Mitri originates from Bari in Italy. As the son of a teacher and a software developer, the foundation for his current field of research was laid early on: Education and Technology. At 19, he founded a web development company and began studying computer science. After a Master's degree in Artificial Intelligence in the Netherlands, employment in the research department of IBM and a doctorate on the topic of multimodal tutors followed. Since 2020, Daniele Di Mitri has been a group leader in the Educational Technologies department of the DIPF.