Prof. Dr. Janice Gobert zu Besuch am IPN – KI-gestützte Unterstützung von Lehrkräften und Schüler*innen in den Naturwissenschaften

Die Abteilung Didaktik der Physik am IPN hatte im September das Vergnügen, Prof. Dr. Janice Gobert von der Rutgers Graduate School of Education (USA) in Kiel willkommen zu heißen.

Während ihres Aufenthalts stellte Prof. Gobert die KI-basierte Lern- und Diagnoseplattform Inq-ITS (Inquiry-Intelligent Tutoring System) sowie das dazugehörige Lehrkräfte-Dashboard Inq-Blotter vor. Beide Systeme wurden entwickelt, um naturwissenschaftliche Kompetenzen von Schüler*innen zu erfassen und zu fördern sowie Lehrkräfte bei der Bewertung und im Unterricht zu entlasten. Prof. Gobert gab Einblicke in die Entwicklung von Inq-ITS, die dabei verwendeten theoretischen Grundlagen und zeigte auf, wie das System in Echtzeit unterstützen kann.

Im Vorfeld ihres Besuchs haben wir mit Prof. Gobert über ihre Forschung und ihre Perspektiven auf KI-gestützte Werkzeuge in der Bildung gesprochen (Interview in Englisch).

Interview:

IPN: What motivated you to develop Inq-ITS and Inq-Blotter, and how do they differ from traditional assessment tools in science education?

Prof. Gobert: There continues to be a great need to support students in learning 21st Century competencies, especially for STEM; for example, 21st century competencies are needed to fill STEM jobs, especially given the increasing technological sophistication of every facet of life. However, many students struggle in STEM; for example, on the last PISA, Germany ranked 22nd and the US ranked 16th. Students need knowledge and skills that go beyond rote knowledge of facts and formulas; rather students need to be able to deeply understand STEM and be able to do the types of tasks that STEM jobs require—in a nutshell, that was the motivation for Inq-ITS, our student platform, which does real time, AI-based assessment of students‘ STEM competencies and supports them also in real time, also using AI, while they learn. Additionally, teachers need to be able to assess these competencies and support their students in learning them, but traditional assessments such as multiple choice tests cannot capture students‘ developing competencies, and open response items are also not good at capturing what students understand, especially if students are writing in their 2nd language - in a nutshell, that was the motivation for Inq-Blotter- to provide teachers real time, AI-driven alerts at a fine-grained level of specificity as to which students need help, what aspect of science inquiry they need help on, and how to help them.

IPN: How do you see AI-based platforms like Inq-ITS changing the role of teachers in the classroom?

Prof. Gobert: Inq-ITS is changing how student learn science competencies because it is giving them real-time, targeted support when student need it! Inq-Blotter is giving teachers real time alerts to guide their instruction to the whole class, to small groups, or to individual students so they don‘t fall behind their peers. Inq-Blotter provides actionable data, alerts, and TIPS (Teacher Inquiry Practice Supports) to guide teachers.

IPN: In your view, what are the biggest challenges in using AI and data-driven tools responsibly in education?

Prof. Gobert: There are a few challenges, but perhaps the one that is a huge problem is the presence of tools like ChatGPT, which if used to replace deep thinking on the part of students, will be detrimental to them—this is the exact opposite of what 21st century skill frameworks are seeking. A second problem is developing people’s understanding of the benefits and risks of different kinds of AI, such as supervised versus unsupervised AI; these are also related to guardrails for the use of AI tools in classrooms.

IPN: What is the purpose of visit to the IPN in Kiel? Are there particular areas of overlap between your work and current projects at the IPN that you are especially interested in exploring?

Prof. Gobert: I gave a talk on my work and presented an extension to the presentation that I gave at the conference in Frankfurt. There are many people working in Kiel whose work overlaps with mine in terms of seeking to deeply improve students‘ STEM learning and teachers‘ pedagogical practices using AI, all of which recognize the science classroom as an ecosystem of learning. The faculty in Kiel recognize that STEM learning has both domain-general and domain-specific facets, e.g. there are separate departments for Physics Education, Chemistry Education, etc. This will, no doubt, lead to deepened understanding of STEM learning across and within specific domains.