Research topic Archives - EduTec Science

Apr162024 by Heather WehnerNo Comments

New Pub: Feedback sources in essay writing: peer-generated or AI-generated feedback?

Artificial Intelligence, Empirical Study, Feedback, Further Education, Journal, Publication

A newly published article discusses the use of peer feedback as a learning strategy, particularly in large classes where teachers face heavy workloads. For complex tasks like writing argumentative essays, peers may struggle to provide high-quality feedback due to the cognitive demands involved. The emergence of Artificial Intelligence (AI) tools, like ChatGPT, raises the question of whether AI can serve as a new feedback source for such tasks. To investigate this, a study compared ChatGPT-generated feedback with peer feedback on argumentative essays written by 74 graduate students from a Dutch university. The study collected essay data, peer feedback and ChatGPT-generated feedback, and then analyzed them using coding schemes. Results showed significant differences between ChatGPT and peer feedback, with ChatGPT offering more descriptive feedback while peers focused on identifying essay problems.…

Apr112024 by Sebastian GombertNo Comments

Team led by Sebastian Gombert wins one of two tracks at BEA 2024 shared task on predicting Item Difficulty and Item Response Time

Artificial Intelligence, Assessment, Award, Computational Psychometrics, Conference, Higher Education, New Pub, Workshop

For standardized exams to be fair and reliable, they must include a diverse range of question difficulties to accurately assess test taker abilities. Additionally, it's crucial to balance the time allotted per question to avoid making the test unnecessarily rushed or sluggish. The goal of this year's BEA shared task (competition) was to build systems which could predict Item Difficulty and Item Response Time for items taken from the United States Medical Licensing Examination (USMLE). EduTec member Sebastian Gombert designed systems which are able to predict both variables simultaneously. These placed first out of 43 for predicting Item Difficulty and fitfth out of 34 for predicting Item Response Time. They use modified versions of established transformer language models in a multitask setup. A corresponding system description paper titled Predicting Item…

Mar272024 by Heather WehnerNo Comments

CORE Roundtable in Munich

Assessment, Critical Online Reasoning, Multimodal Learning Analytics, Project, Project meeting

Hendrik Drachsler, Sebastian Gombert and Gianluca Romano participated at the Roundtable in Munich for the CORE project (Critical Online Reasoning in Higher Education) from 04.03.-05.03.2024. In those two days, our team had the chance to recapitulate on how the infrastructure stood strong during the first survey from December 2023 to February 2024, and pave the way for next steps and surveys. In summary, the infrastructure performed well. It dealt with approximately 10Mio. requests per seconds and the majority of hurdles participants reported were out of our authority. Requests from participants were dealt with quickly in a few days on average. Even though we are proud of our achievements there is still a lot to be done for future surveys. For the agenda we split into smaller groups, each of us…

Mar202024 by Heather WehnerNo Comments

Didacta Presentation in Cologne: Key Questions to Find Fitting AI Solutions For Student Feedback

Artificial Intelligence, Feedback, General education, Invited talk, School

What are the goals of AI in education for student feedback? How can teachers make sure that their AI-assisted feedback goes beyond simple right/wrong statements and instead provides not only correct solutions, but also possibilities for improvement, hints on competence development and effective learning strategies? To find a fitting AI solution, there are key questions one should know to ask in advance. These questions were outlined by Hendrik Drachsler in his presentation at the didacta 2024 in Cologne on 20.02.2024 titled “Ihr KI-Anbieter-Test - 3 Schlüsselfragen die Sie kennen sollten.” The key questions: Question 1: What indicators does your AI product use to analyze learning outcomes? --> Look for AI products that provide relevant indicators for measuring learning progress and skills acquisition. These indicators are important to accurately assess learning…

Mar192024 by Heather WehnerNo Comments

Podcast “Sitzenbleiben” About AI in Education

Artificial Intelligence, General education, School

In a special edition of the DIPF podcast "Sitzenbleiben", Kai Maaz (Executive Director of DIPF) and Hendrik Drachsler discuss the potentials and hurdles of integrating artificial intelligence (AI) into education. The conversation surrounding AI's capabilities and constraints has been rapidly gaining traction. Especially the emergence of large language models like ChatGPT has revolutionized the text creation process, which showcase the vast opportunities that AI technologies can bring to the education sector. Questions abound regarding AI's impact on education: How can AI impact tasks such as homework or exams? How can AI enrich lessons? Which data protection issues need to be considered? How can students and educators receive adequate support in navigating AI's integration into learning environments? These questions serve as focal points which are examined in this newly available podcast…

Mar152024 by Heather WehnerNo Comments

New Pub: AI program doesn’t make kids better at math, but it makes them more independent

Assessment, Journal, Publication, School

Students who receive math tutoring from an artificial intelligence (AI) program perform no better than students who are taught by a "real" teacher. These students do, however, need less help learning. This is the conclusion of Rashmi Khazanchi from the Open University of the Netherlands together with Hendrik Drachsler and Daniele Di Mitri. Math Lessons with AI The researchers examined the effectiveness of the Assessment and Learning in Knowledge Spaces (ALEKS) tutoring program, called Intelligent Tutoring Systems (ITS). Previous studies have shown that students learn math better using software than traditional teaching methods. Previous studies on ALEKS have also shown that, thanks to this program, students memorize more knowledge, perform better, experience more engagement in mathematics and drop out less. The advantage of an ITS like ALEKS is that it…

Feb222024 by Daniele Di MitriNo Comments

The future is multimodal – podcast from AI_ducation

Artificial Intelligence, Higher Education

Some months ago, Alessandra Antonaci and Filippo Chiarello, members of the Encore project, invited Daniele Di Mitri to discuss his research in multimodal AI learning analytics. The result of this discussion is in the podcast below.

Jan312024 by Heather WehnerNo Comments

SWK-Talk: Large Language Models and their potential in the education system

Artificial Intelligence, Event, School

In the SWK Talk Special "Large Language Models and their potential in the education system" on 18.01.2024, the SWK (Standing Conference of the Ministers of Education and Cultural Affairs) presented its impulse paper on Large Language Models. For the impulse paper, the SWK consulted external experts, including members of the EduTec Team, on teaching and learning with AI and LLM. The aim was to contribute to the current debate on the potential of LLM in the education system. The key conclusion is that the German education system currently faces the task of trying to utilize the potentials of generative AI technologies such as LLM, while at the same time recognizing their limitations and finding a way to responsibly deal with their restrictions. The paper also emphasizes the importance of a…

Jan292024 by Sebastian GombertNo Comments

New Pub: From the Automated Assessment of Student Essay Content to Highly Informative Feedback: a Case Study

Artificial Intelligence, Assessment, Computational Psychometrics, Empirical Study, Feedback, Higher Education, Journal, Publication, Special Issue, Technical paper

How can we give students highly informative feedback on their essays using natural language processing? In our new paper, led by Sebastian Gombert, we present a case study on using GBERT and T5 models to generate feedback for educational psychology students. In this paper: ➡ We implemented a two-step pipeline that segments the essays and predicts codes from the segments. The codes are used to generate feedback texts informing the students about the correctness of their solutions and the content areas they need to improve. ➡ We used 689 manually labelled essays as training data for our models. We compared GBERT, T5, and bag-of-words baselines for both steps. The results showed that the transformer-based models outperformed the baselines in both steps. ➡ We evaluated the feedback with a learner cohort…

Jan202024 by Daniele Di MitriNo Comments

Interview: A Controlled Way to Better Teaching and Learning with AI

Artificial Intelligence, Press

DIPFblog Interview with Dr. Daniele Di Mitri about the project "HyTea – Model for Hybrid Teaching“ Artificial intelligence (AI) has the potential to support teaching and learning in many automated ways. However, the contributions of the new technology do not always match the expectations and values of human users. The research and development project „HyTea – Model for Hybrid Teaching“ is investigating how this problem can be addressed. In the interview, project leader Dr. Daniele Di Mitri explains in more detail the project and how he and his team are proceeding. The interview on the DIPFblog – in English and German