The instructors made better evaluations of the learning problems after receiving feedback from AI.

The instructors made better evaluations of the learning problems after receiving feedback from AI.

An example of a different concept, created by AI for a teacher’s evaluation of a student with learning difficulties (German, with translation notes). Photo: Michael Sailer, LMU-Munich

An experiment in which trained teachers are identified to identify students with learning difficulties that are ‘marked’ by intellectual property.

The study, with 178 lecturers in Germany, was conducted by a research team led by faculty at the University of Cambridge and Ludwig-Maximilians-Universität München (LMU Munich). It provides some of the first evidence that artificial intelligence (AI) can enhance the ‘diagnostic thinking’ process: the ability to collect and evaluate evidence about a student, and draw appropriate conclusions. so that appropriate support can be provided.

During the trial, instructors were asked to evaluate six students who were compared with learning difficulties. They are provided with samples of their schoolwork, as well as other information such as work history and records of conversations with parents. They then need to determine if each student has learning disabilities such as dyslexia or Attention Deficit Hyperactivity Disorder (ADHD), and explain their cause.

Shortly after submitting their responses, half of the trainers found a prototype ‘expert solution’, first written by a qualified professional, to compare with their own. This is a common pattern that students experience outside of the classes taught. Others have had AI -generated ideas, which have highlighted relevant parts of their solution and improved flags.

After completing six preparatory exercises, the instructors took two similar experiments – this time without expectation. The trials were reviewed by the researchers, who evaluated their ‘diagnostic accuracy’ (whether the learners had a clear idea of ​​the symptoms of dyslexia or ADHD), and their diagnostic thinking: how good their use of available information to make this decision.

The average score for diagnostic thinking among learners who experienced AI thinking during the first six exercises was 10 percent higher than those who practiced with previously documented technical solutions.

The reason for this is the ‘adaptive’ nature of AI. Because of looking at the teachers’ own performance, rather than asking them to compare it with a technical authority, the researchers felt that the concept was clearer. Thus, there is no evidence that this type of AI improves the one -to -one approach from a human educator or senior educator, but researchers show that there is no way to get close support from educators for practice. again. training, more people in the big classes.

The study is part of a research program within the Cambridge LMU Strategic Partnership. The AI ​​was developed with support from a team at Darmstadt Technical University.

Riikka Hofmann, Associate Professor in the Faculty of Education, University of Cambridge, said: “Teachers have an important role to play in recognizing the symptoms and problems of teaching students and directing them to The level of personal guidance teachers in German classes differs from the UK, but in both cases AI can provide a new level of personal concepts to help them develop these important skills. “

Dr. Michael Sailer, from LMU Munich, said: “Of course, we don’t argue that AI replaces educators: new educators always need expert leadership in recognizing learning problems in the first place. expect these learners to focus on what they need to learn.

The study used a natural language editing system: an artificial neural network that could analyze human language and identify certain words, ideas, hypotheses or evaluations in students ’text.

It was performed using the responses of a first group of first responders in a similar exercise. By sorting and counting these responses, the team instructed the system to identify the presence or absence of important factors in the answers given by the instructors during the session. the trial. The system then selects the pre -written blocks of text to provide the appropriate feedback.

During the preparatory exercises and subsequent activities, participants were asked to perform individually or in pairs. Those who performed the same and received technical solutions during the preparatory training accounted for, on average, 33% for their diagnostic thinking during the follow -up sessions. In comparison, those with the AI ​​idea scored 43%. Similarly, the average number of instructors doing doubles was 35% if they had the technical result, but 45% if they had received support from the AI.

Training with AI had no significant effect on their ability to accurately visualize simulated students. Instead, it was changed by helping teachers cut out the various sources of information they were asked to read, and provide more detailed descriptions of learning problems. This is the key skill that most teachers in the classroom need: the task of seeing students fall into special education teachers, psychologists, and the medical profession. Teachers need to communicate and show their observations to professionals where they are concerned, to help students access appropriate support.

The extent to which AI can be used extensively to support the basic thinking skills of teachers is a broad question, but the research team hopes to redesign the study to find practices that have worked best. in this case, and evaluate this broad possibility.

Frank Fischer, Professor of Educational and Educational Psychology at LMU Munich, said: “In large -scale training programs, it is common in schools such as teacher training or health education, to use AI to support learning simulation can have real value.Implementing complex natural language tools for this purpose takes time and effort, but if it helps in improving thinking skills of future cohorts will probably prove the value of money. “

The research is published in Teaching and learning.


Computer games in the classroom: Educational success is at the core


More information:
Simulation-based simulations from Artificial Neural Networks facilitate the training of teachers before conducting Simulation-based learning. Teaching and learning (2022). DOI: 10.1016 / j.learninstruc.2022.101620

Presented by Cambridge University

Directions: Teachers have made better evaluations on learning problems after receiving feedback from AI (2022, April 11) downloaded on 11 April 2022 from https:/ /phys.org/news/2022-04-trainee-teachers-sharper-difficulties-feedback.html

This document is subject to copyright. Except for appropriate action for the purpose of personal inquiry or research, no piece may be reproduced without permission. Information is provided for informational purposes only.

Related Posts

Leave a Reply

Your email address will not be published.