How does technology help student-teacher interactions

You can see a young teacher explaining something to his tutor. Nervous display needles flicker up and down below the screen. They show the teacher's emotions in the online classroom discussion. At least what the machine thinks it identifies as anger, sadness, or joy. A second camera follows the student - and also measures his feelings.

The experiment is a novelty in Germany. The teacher works for "GoStudent", a provider of online tutoring. GoStudent is one of the few education companies that aggressively use artificial intelligence (AI) for learning. "With the help of AI, we can make lessons more transparent," says founder Felix Ohswald.

Whether AI is really capable of improving teaching is at least controversial. For some, its use is a matter of course, for others the technology plunges into an ethical dichotomy. What is certain: Germany is at the beginning of the era of artificial intelligence. And with Corona, the data age has also begun in schools. The access of young people to school clouds, for example, recorded greater growth last year than the use of Tiktok - after all, the youth medium with the most aggressive growth worldwide. Now there is a massive amount of data. This enables the use of artificial intelligence. Even the first state schools are preparing. But everyone is very quiet, because AI is frowned upon in education. According to a survey, 70 percent of Germans are strictly against "digitally recording the learning progress of children and having it evaluated by algorithms".

These scruples do not bother the 26-year-old Austrian Ohswald. He is certain that he can use AI to find out "why lessons with teacher A go better than teacher B". He claims to have deciphered the mysterious bond of successful interaction between student and teacher. His helpers: the camera eye and "iMotions", a software for emotion recognition. Your algorithm is trained on millions of faces.

The findings so far are rather trivial, measured against one's own demands: "If an emotion is shown by the tutor: in, the same emotion is also present in the student: in at the same time." The students, it is also said, "do not experience high emotional intensity". In German: The tutoring is often rather boring - and not stimulating.

"But it is important for individuals to have a rest."

Felicitas Macgilchrist is critical of such use of artificial intelligence. The Göttingen researcher, who examines data-driven learning software in the "Datafied" project, has something against the permanent camera observation of children. "Then there is no more invisibility," she says. "Not a moment when the student is not being observed. With this technology, students are constantly being data. But it is important for individuals to have a break in which they do not feel supervised. This is central to the feeling of autonomy."

At least GoStudent is trying something out. So far, digitally available data has hardly been used in the German school system. England skipped the Abitur exams last year and calculated the grades using an algorithm. That caused a lot of resentment, but the decisive factor is that it was possible because the British school authority Ofqual is collecting massive amounts of data on students. Saxony, on the other hand, had no data when the math exam last year was unusually bad. So the school minister had to make random phone calls to his high schools to find out where the problem was.

Education start-ups, on the other hand, with Corona taking center stage, are closely following the hundreds of thousands of learners on their platforms. However, they are also familiar with the harsh rejection of the AI ​​instrument in Germany. Hence, they are careful. The quotable part of conversations usually ends abruptly when it comes to AI.

The learning management system (LMS) Scobees, for example, which was developed with two reform-pedagogical schools in Marburg and Munich and used by 650 schools across Germany, could in future automatically offer students recommendations for learning with the help of AI. "Many teachers ask about this extension of the LMS in order to have more time to support the children," reports Scobees founder Annie Dörfle. However, the tool has not yet been programmed.

At the young university school in Dresden you are already one step further. Its founder and scientific director, Anke Langner, is a learning reformer dyed in the wool. "My idea of ​​the development process is that the meaningfulness for learning always comes from the student himself," says Langner. That's why she tries with the teachers of the school attached to the university to show children individual learning paths. Soon this will also be done with the help of an algorithm that the teachers have been diligently feeding with learning tracks since the school was founded.

Anke Langner, who accompanies the school as a professor of inclusive education at the TU Dresden, does not see herself "as a positivist who wants to use AI in five years, but as a critical citizen". Nonetheless, as a scientist, she wants to "explore the potential of using AI to observe the learning behavior of students herself". In doing so, she will also examine the behavior of the teachers. She and her team want to know whether teachers rely solely on AI to provide learning materials and assignments. "If the algorithm controls learning with its recommendations," says Langner, "I would definitely take a step back."

Wait a year for the students to finish their learning? A learning portal promises: We can do it faster

You can see what power AI can have in learning at Simpleclub. The flashy math videos by students Nico Schork and Alex Giesecke in the 2010s have turned into a serious learning portal. One million registered users continuously provide data. Giesecke emphasizes that anyone who learns at Simpleclub is not served any ready-made bits of knowledge. "Depending on how the student gets along, he'll receive an individually tailored program from us - updated to reflect his or her current level of learning."

Giesecke and Schork have offered the Federal Ministry of Education to carry out an exemplary study of learning levels for their users. So far there have been many complaints about the learning deficits of the "Generation Corona", but little is really known. The first results of the learning status measurement planned for autumn by the renowned Bamberg Institute for Educational Research are expected in one year at the earliest. "Surveying the learning status is our daily bread," says Giesecke on the other hand. "We can do that - and politics needs it."

That is why the two founders are ready to expand their offering. All Schoolchildren of certain age groups can work on selected math, German and science tasks in the Simpleclub app. The "Jarvis" technology could then show an anonymized learning status for the collective Corona class. Nobody would be obliged to subscribe to Simpleclub. It would just be a quick X-ray of the Corona student patient that everyone is worried about right now.

Giesecke and Schork would even go a step further: they could not only describe the group of students with deficits according to size and subjects, but also say exactly which student is where. Then the ministers for family and education, Franziska Giffey and Anja Karliczek, who are currently fighting for the addressees of their two billion corona catch-up program, would no longer poke in the fog. You would then know which students need help and where.

Giesecke and Schork would like to prove that artificial intelligence can be useful in education. They have not yet received an answer from the ministry.