The siblings Akshaya and Raksan Mylvaganam learn with Anna, the AI tutor, mathematics.Image assembly: Andrea Zahler

Two francs per hour, available around the clock and amazingly patient: a AI chat bot developed by ETH students replaces the classic tutor. This could be a real progress, especially for children from families with little money – if the motivation is right.

25.04.2025, 22:2225.04.2025, 22:22

Lisa Kwasny / ch media

In a high -rise building in Chur, embedded between the Graubünden mountains, the twins are sitting in front of the laptop. A exercise sheet with math tasks is displayed on the screen, and her tutor Anna can also be seen in a tile. Anna wears a top that looks like a space driver from a science fiction film.

Anna says “Addition” and “Division” with an English accent, but that doesn’t bother the children. Sometimes Anna does not speak in whole sentences, then the two ask her to repeat the sentence. Basically, Akshaya and Raksan are very patient with their tutor.

But sometimes this thread of patience is also tearing: “seventeen,” says Raksan almost desperately. “Do you mean seventh?” Asks Anna. “No, seventeen!” Calls Akshaya into the microphone. “Seventeen?” Asks Anna, “What do you want to do with the seventeen?”. “The answer is seventeen,” says Raksan visibly annoyed.

“I think you have to restart the program, it has listened to our conversation in the background and is confused,” says Friedrich Wicke. Raksan invites the window to new and Anna reappears. This time she understands the answer right away.

Friedrich Wicke developed the AI tutoring together with Gero Embser.Image: CH Media

Wicke is ETH master student and developed Anna together with his business partner Gero Embser. Because Anna is not an ordinary tutor, she is a AI avatar. Raksan and Akshaya are part of around a hundred children who already use the product.

Father was critical, children were thrilled

Table of Contents

ETH Zurich Students Create AI Tutor: Revolutionizing Learning?

Akshaya and Raksan, both eleven years old, were always good at school. But in fifth grade, math lessons became more demanding and their grades worse. The father Ramesh Mylvaganam could not help his children. He works from early to late as a production manager for plastic beam materials in Chur.

In Sri Lanka, he was at school for ten years before he fled to Switzerland before the war. But with the German language he still has trouble, like his wife Jaya. In addition, his children find it anyway, the parents explained everything wrong. So a tutor was needed.

“Actually, I wanted a real person as a tutor,” says Mylvaganam. But these cost 25 to 130 francs per hour. These costs could hardly be borne for the sole earner. On the Internet he came across «tutor.new», the start-up of ETH students. With the AI tutor, the hour only costs two francs.

The father Ramesh Mylvaganam first wanted a human tutor. In the meantime, however, he is satisfied with the AI.Image: Andrea Zahler

Mylvaganam agreed to make a test run, but was critical. The children convinced her father to try Anna for a while. Now they have been working with the AI teacher for three months and although she still has some potential for improvement, the father is now also satisfied with the solution.

AI tutoring requires high motivation to learn

Is that the future? Are people being displaced by artificial intelligence in school lessons? “So far, AI has not been able to replace teachers broadly,” says Dominic Hassler, expert for AI and lecturer at the Center for Vocational and Adult Education at the PH Zurich. “A prerequisite for tutoring with AI is self -motivation and the competence to learn independently. If you bring this with you, you often don’t need tutoring, neither from AI nor of a person. »

If you practice AI, you need a high level of motivation, says AI expert and PHZ lecturer Dominic Hassler.Image: CH Media

What does it look like at Akshaya and Raksan? Is it difficult to motivate yourself for AI tutoring? Akshaya nods shyly, the mother grins. She sometimes has to persuade the children to sit down to practice, her husband says with a look at the two.

«I also don’t think AI can replace a really good teacher. But in reality, not all children have access to good tutoring, »says Wicke. Good tutoring is expensive, not everyone could afford it. Therefore, the app costs only two francs per hour, which is just above the cost of the server and the operation of the AI models for one hour. The operators are even down with the price, initially he was six francs.

“Of course, we could earn a lot more money as a well -trained computer scientist, but the project also has a political dimension,” says Wicke. Parents are usually busy. Not only to understand subjects like math, but also to explain, could be a challenge – this makes it difficult to make it difficult to know.

Through the AI teacher, the two want to contribute to educational justice. And the offer is growing: On request, Anna no longer only provides tutoring at primary school level, but also gives German lessons for parents.

Advantage: AI can explain a solution

Despite criticism, Hassler also sees the advantages of AI in school lessons: “If you could not solve a partial task in the homework, you just couldn’t get any further,” he says. «Now AI can almost always explain the solution. This is a huge opportunity for the motivated students. »

Even if the relationship with the teacher is problematic or the child does not understand the explanations of the teacher, but the explanations of the AI can help, a AI can help can help.

Akshaya and Raksan are annoying when the AI does not understand them, but overall they find the program good.Image: CH Media

In fact, Anna can explain the solution. And she has even more advantages: she can regularly create new tasks and remembers which topics the children still have to practice. In contrast to exercise sheets on paper, Akshaya and Raksan cannot simply check the solution. If you give the wrong answer, the AI says in a nice voice “try it again”.

Then the two whisper their considerations, quietly so that Anna’s language input is not activated. If you don’t get ahead, ask the AI for help. And Anna explains the calculation path step by step. Sometimes she speaks a little chopped, sometimes too quickly. But as one speaks correctly, Anna can still learn. (aargauerzeitung.ch)

You might also be interested:

With an update except for the series, Apple closes two dangerous security gaps. Three popular devices from the group are affected.

Apple has released an emergency patch for its devices, which removes two critical zero-day weak points in the operating systems, the company reports. The security gaps have already been used for targeted attacks by hackers, which is why Apple offers the update except for the series.

The weak points enable attackers to carry out targeted and “extremely sophisticated attacks” on individual Apple users, as the manufacturer said. Details are not known, but for “usually these are commercial spyware that is used by government agencies against target persons”, Write the Techportal Heise.

date:2025-04-25 20:22:00

ETH Zurich Students Create AI Tutor: Revolutionizing Learning?

A team of bright students at ETH Zurich, a leading science and technology university, have developed a cutting-edge AI tutoring system poised to transform the educational landscape. This innovative project demonstrates the potential of artificial intelligence in education to personalize learning experiences and address the growing demand for effective, accessible educational resources. The AI tutor promises to provide students with customized support, immediate feedback, and a flexible learning surroundings, ultimately enhancing their understanding and academic performance.

How Does this AI Tutoring System Work?

The ETH Zurich team’s AI tutor is built upon advanced machine learning algorithms, natural language processing (NLP), and educational data analytics. Here’s a breakdown of its core functionalities:

Personalized Learning Paths: The AI assesses each student’s individual learning style, strengths, and weaknesses through initial assessments and continuous monitoring of their progress. Based on this data, it creates a customized learning path tailored to their specific needs. This personalized approach ensures that students focus on areas where they need the most support, optimizing their learning experience and accelerating their progress.
Adaptive Questioning: The system employs adaptive questioning techniques, adjusting the difficulty level of questions based on the student’s responses. If a student answers correctly, the difficulty increases. If they struggle, the system provides simpler questions and targeted explanations to reinforce their understanding. This adaptive nature keeps students engaged and challenged without overwhelming them.
Instant Feedback and Explanations: Unlike conventional learning methods where feedback might be delayed, the AI tutor provides instant feedback on every answer. This immediate response helps students identify and correct mistakes in real-time, strengthening their understanding of the concepts. Additionally, the system offers detailed explanations for both correct and incorrect answers, providing a deeper insight into the underlying principles.
24/7 Availability: One of the most significant advantages of an AI tutoring system is its round-the-clock availability. Students can access the tutor anytime, anywhere, allowing them to learn at their own pace and on their own schedule. This flexibility is especially beneficial for students with busy schedules or those who prefer to study outside of traditional classroom hours.
Data-Driven Insights for Educators: The AI tutor collects and analyzes vast amounts of data on student performance, providing educators with valuable insights into student learning patterns. This data can be used to identify areas where students are struggling, allowing teachers to adjust their curriculum and teaching methods to better meet the needs of their students.

Benefits of AI in Education: Beyond Traditional Tutoring

The potential benefits of this AI tutoring system, and AI in education in general, extend far beyond traditional tutoring methods. Here’s a closer look at the advantages:

Improved Student Outcomes: By providing personalized guidance and targeted support, the AI tutor can considerably improve student outcomes. Studies have shown that students who use AI-powered learning tools demonstrate higher levels of engagement, better understanding of concepts, and improved grades.
Increased Accessibility: AI tutoring can make quality education more accessible to students in remote areas or those facing financial constraints. by providing affordable, personalized learning experiences, AI can definitely help bridge the achievement gap and ensure that all students have the opportunity to succeed.
Personalized Learning Experience: Perhaps the most significant benefit is the personalized learning experience that AI offers.By adapting to each student’s individual needs and learning style, the AI tutor can create a more engaging and effective learning environment, leading to better retention and a deeper understanding of the material.
Reduced Teacher Workload: While not intended to replace teachers, AI tutoring can definitely help reduce their workload by automating tasks such as grading and providing individualized feedback. This allows teachers to focus on more strategic activities, such as curriculum development and personalized student support.
Enhanced Engagement: The interactive nature of AI tutoring systems can make learning more engaging and enjoyable for students. By incorporating gamification elements and providing immediate feedback, AI can help keep students motivated and on track.

Practical Applications of ETH’s AI Tutor

The AI tutoring system developed by ETH Zurich students has the potential to be applied in a wide range of educational settings. Here are a few examples:

K-12 Education: The AI tutor can be used to supplement classroom instruction in subjects like math, science, and language arts. It can provide students with personalized support and practice opportunities, helping them master key concepts and improve their academic performance.
Higher Education: In colleges and universities, the AI tutor can be used to provide students with additional support in challenging courses. It can help students review course material, prepare for exams, and improve their understanding of complex concepts.
Test Readiness: the AI tutor can be used to help students prepare for standardized tests, such as the SAT, ACT, and GRE. It can provide students with practice questions, personalized feedback, and test-taking strategies to help them achieve their best possible score.
Professional Development: AI tutoring can also be used to support professional development initiatives. It can provide employees with personalized training and skill-building opportunities, helping them stay up-to-date with the latest industry trends and technologies.
Language Learning: AI-powered language learning platforms can offer personalized lessons, pronunciation practice, and real-time feedback, making language acquisition more efficient and engaging.

Challenges and Considerations

While the potential of AI tutoring is enormous, there are also several challenges and considerations that need to be addressed:

Data Privacy and Security: AI tutoring systems collect vast amounts of data on student performance, raising concerns about data privacy and security. It’s crucial to ensure that this data is protected and used responsibly.
Bias and Fairness: AI algorithms can be biased if they are trained on biased data. It’s important to ensure that AI tutoring systems are fair and equitable, providing all students with equal opportunities to succeed.
Ethical Considerations: The use of AI in education raises ethical considerations about the role of technology in learning. It’s important to ensure that AI is used to enhance, not replace, human interaction and critical thinking skills.
Implementation Costs: Implementing AI tutoring systems can be expensive, requiring investments in hardware, software, and training. It’s critically important to carefully consider the costs and benefits of AI tutoring before making a decision to implement it.
Teacher Training and Support: Teachers need to be properly trained on how to use AI tutoring systems effectively. They also need ongoing support to integrate AI into their teaching practices and address any challenges that may arise.

Case Studies: AI Tutoring in Action

While still relatively new, AI tutoring is already showing promising results in various case studies. Thes examples illustrate how AI can be effectively implemented in different educational settings:

Carnegie Learning’s MATHia: This AI-powered math tutor has been shown to improve student outcomes in algebra and geometry. It provides personalized feedback, adaptive questioning, and real-time data insights for teachers.Studies have indicated improved test scores and overall comprehension among students using MATHia compared to traditional methods.
Duolingo: While primarily a language learning app, Duolingo employs AI to personalize lessons, adapt to individual learning speeds, and provide immediate feedback on pronunciation and grammar. Its effectiveness has been demonstrated in various studies showing significant gains in language proficiency among users.
Khan Academy: khan Academy uses AI-driven practice exercises and personalized learning dashboards to help students master a variety of subjects. Its adaptive platform adjusts to each student’s skill level, providing targeted support and reinforcing concepts as needed.

first-Hand Experience: A Student’s Perspective

To understand the real-world impact of AI tutoring, let’s consider a hypothetical student, Sarah, who used an AI tutor to prepare for her calculus exam:

Sarah, a struggling calculus student, felt overwhelmed by the complex concepts and formulas. She decided to try an AI tutoring platform recommended by her professor. Initially skeptical, Sarah quickly discovered the benefits of personalized learning. The AI tutor assessed her strengths and weaknesses, creating a customized study plan that focused on areas where she needed the most help.

The adaptive questioning feature challenged her without overwhelming her, and the instant feedback helped her identify and correct mistakes in real-time. Sarah found the explanations for both correct and incorrect answers particularly helpful, as they provided her with a deeper understanding of the underlying principles. The 24/7 availability of the AI tutor allowed her to study at her own pace and on her own schedule, fitting her learning around her other commitments.

As Sarah’s confidence grew, so did her understanding of calculus. When the day of the exam finally arrived, Sarah felt prepared and confident. She aced the test, thanks to the personalized support and targeted practice she received from the AI tutor. Her success motivated her to continue using AI tutoring tools in other subjects, transforming her approach to learning and improving her overall academic performance.

Practical Tips for Implementing AI Tutoring

If you’re considering implementing AI tutoring in yoru school or district, here are some practical tips to help you get started:

Start Small: Begin by piloting AI tutoring in a specific subject or grade level to assess its effectiveness and identify any challenges.
Provide Teacher Training: Ensure that teachers are properly trained on how to use AI tutoring systems and integrate them into their teaching practices.
Communicate with Parents: Keep parents informed about the benefits of AI tutoring and address any concerns they may have.
Monitor Student Progress: Track student progress regularly to assess the impact of AI tutoring and make adjustments as needed.
Choose the Right Platform: Research and select an AI tutoring platform that aligns with your curriculum and meets the needs of your students.
Focus on Blended Learning: Remember that AI tutoring should complement, not replace, traditional teaching methods.Focus on creating a blended learning environment that combines the best of both worlds.

Example AI Features
Feature	Description	Benefit
Adaptive Learning	Adjusts difficulty based on student performance.	Personalized learning experience.
Instant Feedback	Provides immediate responses to answers.	Reinforces learning and corrects mistakes.
Personalized Reports	Generates student specific data insights.	Educators gain insight on each student’s educational needs

The Future of AI in Education

The AI tutoring system developed by ETH Zurich students offers a glimpse into the future of education. As AI technology continues to advance, we can expect to see even more innovative applications of AI in the classroom. From personalized learning experiences to automated grading and feedback, AI has the potential to transform education and empower students to achieve their full potential. The continued development and refinement of artificial intelligence in education will lead to a more equitable and effective learning environment for all.

ETH students develop AI tutoring-it can