St Petersburg University sociologists: university students need to develop a new mindset for working with AI


The image is generated by the Kandinsky neural network
The image is generated by the Kandinsky neural network

The research findings are published in the peer-reviewed journal Vysshee Obrazovanie v Rossii (Higher Education in Russia).

Today, generative neural networks are being introduced in almost every sphere of human life, including education. In a matter of seconds, these neural networks can generate various types of information goods — from texts and images to music and speech — based on a given set of data. For example, in some Russian universities, AI tools are used to design simple learning tasks or to analyse and forecast students’ academic performance. Students, on the other hand, can use neural networks to edit their essays and even to write their graduation projects.

That said, the potential for applying such technologies in higher education is even wider. Generative neural networks can help in learning a foreign language, providing the opportunity to practise speaking skills and learn new vocabulary. They can also generate topics, ideas, and outlines for essays and research papers. Finally, such algorithms can be employed to evaluate other works created using AI-based technologies.

St Petersburg University sociologists Andrey Rezaev and Natalia Tregubova analysed a number of publications on the use of AI-based text generators (ChatGPTs) in higher education, produced by researchers from different countries. They identified five key areas where, according to the researchers, artificial intelligence is likely to be used more frequently. These areas include: higher education accreditation and licencing (for automated generation of certificates and reports); student enrolment (for preparing answers to recurring questions from applicants); teaching and learning (to assist in developing curricula and teaching materials); evaluation of the performance of higher education institutions; and creating personalised learning paths for students.

The researchers drew particular attention to the practice of developing and teaching the so-called computational thinking. Andrey Rezaev and Natalia Tregubova stressed that young professionals would need to understand the logic and specifics of human — computer/algorithm interaction, especially in situations when a technical failure occurred.

‘The main goals of such innovations are: improving student performance; making education more accessible, including life-long education; and ensuring that breakthrough technologies are used in an ethical and responsible way,’ said Natalia Tregubova.

It is important that we realise that despite the widespread use of neural networks in many areas of our lives, the responsibility for using AI-based technologies wisely lies with the individual.

Natalia Tregubova, Associate Professor in the Department of Comparative Sociology at St Petersburg University, Candidate of Sociology

Natalia Tregubova added that the active and widespread use of artificial intelligence is another reason to revise the existing technology for assessment of students’ learning outcomes in education and, perhaps, to supplement it with new assessment criteria that could be used by teachers to evaluate students’ written and oral responses.

‘Such approach can be illustrated by the scoring system in figure skating, where different points are awarded separately for technical merit and for artistic performance. There is also a grading practice for students’ essays and summaries in secondary education: one mark is given for content and style, and the other — for grammar and vocabulary. Thus, the skill of using AI tools in preparing a research paper may become a new assessment criterion alongside other criteria, such as research relevance and paper format,’ explained Natalia Tregubova.

The current study is conducted by the researchers at St Petersburg University as part of the project ‘From Artificial Intelligence to “Artificial Sociality”: Everyday Life in Digital Society at the Intersection of Technological and Social Transformations’, supported by the Russian Science Foundation.

Nonetheless, as the researchers stress, we cannot ignore the issues that will inevitably arise when artificial intelligence is actively used in education. The potential issues with generative AI include: a threat to academic integrity; plagiarism and its detection; AI-generated misinformation; superficiality of AI-generated texts; and algorithmic bias. Indeed, all stakeholders in education need to join efforts to deal with these issues in order to establish the necessary boundaries for the use of generative neural networks. Natalia Tregubova pointed out that AI might well become a good assistant for a teacher; yet, it could never replace human teachers completely.

‘I think that in the very near future, artificial intelligence will be able to do the job of an instructor; that is, it will take over the functions for transferring information and checking its assimilation. However, AI will never be able to be a supervisor. A supervisor acts as a guide, allowing students and supervisees to see the issue, the problem (and eventually themselves) in a different way, from another perspective. Artificial intelligence can never bring other perspectives. AI does not have its own view of life and the world, it only summarises the viewpoints and actions of the people who have created the data it learns from,’ Natalia Tregubova emphasised.


Source link

Similar Posts