AI Assistant Grades MPEI Students’ Work and Creates Test Questions
At the Moscow Power Engineering Institute (MPEI), an AI algorithm has learned to grade handwritten student calculations and explain in just 90 seconds exactly where a solution went off track.

The Exam-Edu MPEI AI assistant has graded more than 6,000 student assignments, providing detailed feedback on each one. In doing so, the tool takes over some of the time-consuming routine work that can be particularly burdensome for faculty.
While the Mistake Is Still Fresh
An AI assistant’s value goes beyond its ability to generate text. This software module can create assignments across more than 40 academic disciplines, recognize handwritten problem solutions and provide detailed explanations. It pinpoints where a student made a mistake and what may have caused it. This is particularly important for students studying technical subjects, where the final answer is only part of the picture. The calculations and the reasoning behind each step matter just as much.
Students receive not only a grade but also an explanation. While the mistake is still fresh in their minds, they can return to the problem and work out what went wrong. MPEI Rector Nikolay Rogalev explains: “AI takes over routine but time-consuming tasks: generating standard materials, grading assignments and checking compliance with academic and technical standards. This allows us to shift the focus toward developing critical thinking skills and solving unconventional problems – exactly what future engineers really need.”

When Grisha AIsaev Finds His Voice
Russian universities are taking different approaches to working with AI, but they are moving in the same direction. They were initially cautious about the technology; now they are bringing it into their day-to-day operations.
For example, in 2023, Moscow City University allowed students to use AI tools when working on their graduation projects, provided they reworked the material generated by the algorithms themselves. Notably, that same year UNESCO published its guidance on the use of generative AI in higher education. A Russian translation followed in 2024.
Also in 2023, Tyumen State University first tested an AI assistant developed by Sberbank. The tool analyzes recordings of classes and evaluates lesson structure, how instructors allocate their time and their communication style. Tyumen State University Rector Ivan Romanchuk explained the decision: “I see this step as one element of our long-term partnership with Sber and as a way to advance our strategic goal of introducing artificial intelligence tools into education.”
In spring 2026, Blagoveshchensk State Pedagogical University gave 750 faculty members and students access to a similar service.
RUDN University has some interesting projects of its own. Its faculty members have access to a virtual assistant that helps analyze lectures, while students can turn to Grisha AIsaev, a chatbot built on Sber’s GigaChat language model. For now, Grisha communicates through a text interface, but its developers plan to give it a visual avatar and a distinctive voice in the future.

Staying Within the Lines
More than 6,000 AI-generated reviews of MPEI students’ work are a significant achievement. But faculty members also wanted to establish clear boundaries for the use of AI tools. Back in summer 2026, MPEI approved a policy governing the use of generative AI technologies. The document states that AI’s primary purpose is to “develop students’ cognitive abilities, critical thinking and creativity, not to do their thinking for them.”
Automated Grading Still Needs Validation
In the near future, Russian universities are likely to move beyond isolated AI experiments and begin using the technology regularly in their teaching. MPEI’s experience already shows that assignments completed by engineering students can be graded automatically. But more in-depth research is needed to establish whether machines can assess this work more accurately than instructors. The goal is not simply to expand the use of these systems, but to make sure they deliver reliable results.
The next challenge for Russian developers is to establish a clear framework in which AI-generated analysis complements instructors’ own assessments. That would help maintain confidence in the fairness of grades while ensuring that algorithms do not displace the development of engineering skills and judgment.









































