What tasks do teachers now perform with AI?

We recently conducted a study on AI technologies in edtech. We studied successful examples of using GPT solutions on different platforms and found out what tasks in teaching are now performed by AI instead of teachers.

In our analysis, we focused on platforms that educate adults. We excluded language schools, since it was important for us to consider solutions relevant to learning a new profession.

All of our findings can be divided into four groups by type of function: feedback, virtual assistant consultations, personalized learning plans, and teacher assistance. We will tell you more about each of the groups using examples.

Feedback

GPT models in educational platforms instantly provide students with feedback: they highlight incorrect answers, help correct the error and improve the result.

Correcting errors in the code. GPT models automatically check your code, report errors, and suggest where and what changes need to be made. For example, if a student chose the wrong character or forgot to close a parenthesis, the model will suggest the correct syntactic option. This feature is used in practical simulators in several courses: the YandexGPT neural network highlights errors in the code and explains them.

Hints in assignments and tests. If a student is having difficulty solving a problem, the AI will explain what steps need to be taken and where to start. In quizzes and tests, he will give a hint at the student’s request.

Detailed feedback on text assignments. Some AI assistants not only check text for errors, but also analyze its content, structure and style. As a result, the student receives an expanded list of comments: strengths, growth points and recommendations for improvement. The model tells you if the work lacks examples or citations are formatted incorrectly.

Personalized recommendations. Students can now upload their homework to the platforms and receive specific recommendations from AI. The model will ask clarifying questions and help you better understand key ideas and complex concepts. For example, a neural network can ask why a student chose this particular algorithm, offer other options and help him figure out the solution step by step.

Consultations with a virtual tutor

Edtech platforms are increasingly introducing “virtual tutors”—AI assistants that simulate live dialogue and help students in the learning process and completing assignments.

Code generation. AI models can not only find and correct errors in existing code, but also generate new code upon request. YandexGPT helps students in development courses automatically create code for functions, classes or data processing in different programming languages. In theory lessons, AI models explain the code so that students better understand its workings and logic.

Working with documentation. A student can upload a file and discuss the contents of the document with the AI assistant – the AI will quickly find the necessary information in the document or explain complex concepts in simple language.

Answers to questions. AI answers standard questions about training programs, helps find the necessary information in materials, and suggests how to solve technical problems with code or setting up tools.

YandexGPT provides Practicum students with highly specialized “consultations”: for example, if a developer has questions about programming languages, he can ask them to the neural network and get answers about the syntax, functions, libraries and other aspects of the language. YandexGPT will tell the marketing student about marketing strategies and tools.

Explaining complex topics. If a student does not understand a topic, he can get an explanation from a neural network. The AI assistant will explain the theory in simple terms, give real-life examples, ask questions to reinforce the material, and share additional resources for learning.

A brief recap of the lesson. To quickly review the material covered, a student can ask the AI assistant to summarize the lesson. The neural network will retell the main ideas and highlight key points. There is such a feature in the Workshop.

Practical training. AI assistants ask students follow-up questions to identify knowledge gaps. Based on the answers, the model provides personalized feedback: it offers additional tasks, explanations and examples. Repeating the cycle, the student gradually consolidates the topic and corrects mistakes.

Individual development plan

Many edtech platforms use GPT models to offer students a personalized learning experience. Such models perform three key functions: analyze results, select individual tasks and formulate a development plan.

Analysis of results. AI analyzes students’ successes: assesses their strengths and weaknesses, level of knowledge and progress in learning.

Personalized tasks. Based on the analysis of the results, the AI offers tests and exercises that correspond to the student’s current level of knowledge and help improve understanding of the topic.

Development plan. The AI constantly updates its answers and hints based on the student's progress. Based on performance data, it generates a list of learning materials that best suit the student’s individual needs.

Help the teacher

AI features are aimed not only at students, but also at teachers. GPT models help save time on routine tasks: checking assignments, identifying plagiarism, fixing problem areas and analyzing student performance. At the same time, the role of the teacher remains – they can adjust the AI feedback and customize the system to suit their tasks.

Checking for plagiarism. AI detects plagiarism and code similarity in student papers. If a student copied part of the text or code without indicating the source, the system will detect this and indicate borrowing in the report.

AI check. GPT models distinguish between texts written by humans and those generated by artificial intelligence. If a student used AI to prepare an essay, the system will alert the teacher.

Extended feedback. AI analyzes how students solve problems, identifies errors and makes recommendations to improve results. The code is evaluated based on three parameters: code design, documentation, and algorithm efficiency. The teacher can always correct or supplement the feedback.

Analysis of results. AI identifies questions that students are having difficulty with and tells the teacher which topics require additional attention. Collects statistics on assessments and generates reports with dashboards, from which you can clearly see how many students from the group have difficulties with a specific topic.

We talk more about how AI models help in work and learning in our free course “YandexGPT for beginners”. And a new module on working with neural networks has appeared in the key courses of the Workshop – they talked about this in more detail in a separate article.