Advantages and disadvantages of neural networks as a means of teaching foreign languages to secondary school students | Статья в журнале «Молодой ученый»

Отправьте статью сегодня! Журнал выйдет 16 ноября, печатный экземпляр отправим 20 ноября.

Опубликовать статью в журнале

Автор:

Научный руководитель:

Рубрика: Педагогика

Опубликовано в Молодой учёный №44 (543) ноябрь 2024 г.

Дата публикации: 31.10.2024

Статья просмотрена: 9 раз

Библиографическое описание:

Тюлеева, С. М. Advantages and disadvantages of neural networks as a means of teaching foreign languages to secondary school students / С. М. Тюлеева. — Текст : непосредственный // Молодой ученый. — 2024. — № 44 (543). — URL: https://moluch.ru/archive/543/118797/ (дата обращения: 07.11.2024).



The article examines the potential advantages and disadvantages of introducing AI into the school educational process.

Keywords: artificial intelligence, school, neural networks, foreign language.

In recent years, the use of artificial intelligence in the educational process has been increasingly discussed in modern education. This concerns the possibility of using AI both in higher education institutions and in secondary schools. The study Artificial Intelligence and English Language Teaching: A Systematic Literature Review, conducted by the British Council in 2023, found a sharp increase in the number of scientific publications on the subject since 2017. The authors attribute this to the widespread use of large language models — LLM, Large Learning Model — and their increasing use in ELT/L, English Language Teaching/Learning. These are language models such as ChatGPT, Google's Bard and Microsoft's Bing Chat [1, p.7].

Researchers also note that along with the growing enthusiasm, there are concerns about the partial or total elimination of traditional pedagogical practices such as lectures and oral explanations by teachers. It is also said that when using neural net-based learning, learners have access to a wide range of vocabulary units, such as a large number of synonyms. On the other hand, we still remember an approach to English language learning such as Special English, developed for the Voice of America radio station, with an average vocabulary size of 1500 words, which ensured that lexemes were repeated and therefore memorised.

Bing Han, in his article Application of Artificial Intelligence in Autonomous English Learning Among College Students, cites the possibility of autonomous learning by students with the help of AI as an undoubted advantage. The author points out that in schools and colleges, the number of hours devoted to language learning may be insufficient, and sometimes there is no individual approach to each learner. At the same time, large language models involve individualised human-to-machine LLMs. The author argues that «traditional teaching methods are no longer able to meet the needs of today's society for the development of complex skills, especially in the field of English language teaching». He believes that «the traditional model of college English teaching should be reformed by integrating computer technology and artificial intelligence technology to increase students' interest in learning and to reduce the teaching load» [2, p.63].

Göçen and Aydemir, in their paper Artificial Intelligence in Education and Schools, speculate on whether artificial intelligence will be able to replace school teachers in the future. Referring to other authoritative authors (Haseski, Humble and Mozelius, etc.), the researchers write that AI will contribute to the individualisation of education, but at the same time it will reduce the workload of teachers. Their view that «AI can provide access to relevant and better learning opportunities for excluded individuals and communities, people with disabilities, refugees, out-of-school youth and those living in isolated communities» is hard to argue with. The authors also mention the futuristic 'dual-teacher' model: «a dual-teacher model with artificial intelligence in terms of individualised education: teachers spend a lot of time on routine and other administrative tasks» [3, p.14].

Cantos and Magayanes agree that «AI can help learners learn a new language quickly and successfully by providing personalised learning, feedback, adaptability and access to resources». At the same time, however, they doubt «that technology will ever completely replace teachers. Teachers have a particular combination of skills and characteristics that AI cannot replicate in the classroom, including emotional intelligence, adaptability, creativity, relationship building and judgement» [4, p.56].

UNESCO's research shows concern about the rapid growth of neural networks and their use outside of regulations designed to protect teachers. UNESCO's Assistant Director-General for Education, Stefania Giannini, points out that there is currently no evidence that generative applications such as ChatGPT improve educational outcomes. There is no doubt that AI has potential for personalised learning, she says. «But we also believe that education is a collective and social endeavour, and schools are a place where children socialise and learn to live together», says Stefania Giannini [5, p.1].

The study also mentions the IMF's concern that around 60 per cent of jobs will be replaced or significantly affected by artificial intelligence in the future. It points out that the application of AI in education will be very different in developed and developing countries. Differences in technological infrastructure, funding, political support and levels of digital literacy need to be taken into account. Third world countries still face fundamental problems, often lacking the basic requirements for technology and quality education, from infrastructure to electricity.

UNESCO calls for red lines to be respected when introducing AI in education. The first of these is the protection of privacy and personal data. Schools must be physically and digitally secure. There should be an age limit of at least 13 years for the use of at least 13 years for the use of AI tools.

Crompton and co-authors in the British Journal of Educational Technology take this discussion further by dividing the people using AI in education into three categories. The first is students. The second is teachers. The third category is managers and administrators involved in assessing learning outcomes and general management of the educational institution. The authors' example of the physical embodiment of AI is interesting, such as the creation of the humanoid robots Eric and Sophie, capable of mimicking facial expressions and body language. The use of such machines provided a more realistic approach to teaching students [6, p.2517].

The authors of the article look at the benefits of machine learning in different aspects: speaking, writing and reading. One of the disadvantages of traditional language courses and school lessons is the disconnection from the English-speaking environment at the end of the lesson. In many experiments, AI has connected learners to the language on demand. In Taiwan, for example, the use of AI in English language learning saw improvements in pronunciation as the neural network used visual spectrograms to illustrate pitch in intonation patterns. Another experiment used Alexa, a virtual assistant from Amazon, as an interlocutor and language coach. «Learners», write the authors of the paper, «found that Alexa was easy to use and promoted meaningful interaction, supported vocabulary acquisition, improved language skills, and provided interesting, enjoyable learning.

Experiments with teaching writing skills were conducted using the artificial intelligence-based programme Grammarly. It was found that «the AI system helped improve or expand vocabulary, especially when learning specialised or unambiguous expressions». Grammarly is a grammar checker, and students who used the app made fewer grammatical errors and wrote with more lexical variation.

The potential of neural networks in English language learning is also being explored by researchers from the CIS. For example, Russian scientists Pak and Kryukova used in their work such AI programmes as GigaChat (a neural network from Sber), Speechling (an American non-profit educational platform), Replika (a US company founded by Russian programmers) and Mondly (a product of the British educational company Pearson). The authors of the study compared the capabilities of the above neural networks and developed a series of exercises for teaching foreign languages using these programmes [7, p.71].

When comparing the capabilities of the applications, aspects such as functionality, usability, security, communication with technical support and cost were taken into account. It was found that GigaChat is not suitable for foreign language learning due to its monolingual nature. Speechling functions as an educational application, it has the ability to teach students with different levels of preparation, as well as a separate section for students planning to take IELTS and TOEFL tests. Replika is a chatbot that allows you to take part in various discussions in English and practice your speaking and writing skills. Mondly also allows you to participate in discussions, but is almost entirely paid.

As a result, the researchers selected Replika as the most affordable and functional programme for further testing with the students. An important advantage of the programme was considered to be its flexibility and the ability of the application to automatically adapt to the language level of each learner.

In the future, authors of such studies should be more careful when choosing platforms for learning a foreign language. Popular AI language platforms such as Duolingo, Babbel, Talkpal, Rosetta Stone, Lingvist and Busuu were overlooked. Duolingo, for example, introduced a language learning model to its functionality a year ago, which allows it to generate online tasks for users, as well as synthesise human speech and use this skill in real-time feedback.

The introduction of AI into the mechanism of online language learning applications is particularly rapid today, and the situation in the language platform market is changing literally every day. The example of Duolingo, which has reduced its staff by ten percent thanks to AI, clearly shows that neural networks make it cheaper to develop new application features and make the level of communication between the application and the user as realistic as possible.

The material by O. Tolstykh and E. Panasenko «The potential of artificial intelligence in language teaching: practical recommendations for teachers» discusses the integration of artificial intelligence technologies into modern language teaching. Among the undeniable advantages of introducing neural networks into the educational process, the researchers point to personalisation: personalised tasks for each student, taking into account their level of readiness; automation of content creation; interactivity and feedback with students; use of services such as interactive chatbots or applications; flexibility and cost-effectiveness [8, p.392].

The authors classify all platforms that use AI in their functionality and are suitable for student learning in schools into the following categories. These are «adaptive learning platforms that use AI algorithms to dynamically adjust learning content and analyse individual student needs and performance; language learning applications such as Duolingo, Babbel and Rosetta Stone; language assessment tools such as TOEFL iBT, Pearson that rely on AI for data processing and feedback; speech and pronunciation recognition tools; AI-based voice generation tools». A separate category includes services that enable the generation of educational content, such as the Twee service, which helps to create exercises in the areas of reading, vocabulary, grammar, listening, speaking and writing.

At the same time, G. Tenitsky, who agrees with the opinion that AI can improve the quality of education in secondary schools, identifies a number of problems that may arise in the process of introducing new technologies. First of all, there are technical problems. Integrating AI into existing educational systems may require the acquisition of sophisticated technical infrastructure. High-quality computing resources necessarily mean trained staff and the latest software. All of this implies significant material investment.

The researcher also believes that «extensive and high-quality learner data will be required for AI systems to work effectively. In reality, however, access to such data may be limited or insufficient to train algorithms». Hence, ethical issues «related to the privacy of learner data, the transparency of algorithms, and the fairness of recommendation and assessment systems» [9, p.429].

Meanwhile, in 2024, experiments to introduce AI technologies have already begun in schools in Kazakhstan. In Astana's School No. 84, for example, teachers are using ChatGPT as a teaching assistant during lessons. Ruslan Karatabanov, vice-president of the I. Altynsarin National Academy of Education, said that «the Ministry of Education and the Ministry of Higher Education and Science are jointly creating a Kazakh artificial intelligence. It is at the stage of development. In the future, within 3–4 years, the first Kazakh artificial intelligence based on the Kazakh language will appear». Apparently we are talking about an application or online platform like ChatGPT, which will generate content in Kazakh. Such efforts can only be welcomed.

In general, when summarising the advantages and disadvantages of using AI in the school educational process, it should be said that neural networks offer many advantages as an auxiliary tool when the process is managed by a foreign language teacher. The teacher uses AI as a tool to speed up the preparation of supplementary materials, audio and video content and test tasks. AI applications can help to automate the checking of completed tasks and the evaluation of test results, as is already the case in the practice of international computer-based language testing. A more individualised approach to learning is also positive, as even within the same class, students may have different levels of preparation. The use of AI programmes will make it possible to continue lessons after school, thus providing the necessary number of lessons if they are not sufficient in the school curriculum. The peculiarity of foreign language learning is that it requires daily lessons, and programmes based on neural networks will make it possible to have almost round-the-clock access to exercises.

However, the disadvantages identified by the authors of the studies are mainly technical. For example, the high cost of developing applications, the time required to train specialists, the process of implementing the technology and, usually, the availability of high-speed Internet. However, as the successful practice of computerisation and digitisation in Kazakhstan shows, the technologies implemented pay for themselves over time, saving time and providing wider coverage of the population. At one time, the country had a programme to eradicate computer illiteracy, which consumed a lot of money, time and human resources. Today, there is no need for such a programme as computer literacy is close to 100 %. It is believed that in time the practice of using AI in the teaching process will become as commonplace and habitual.

References:

  1. Artificial Intelligence and English Language Teaching: A Systematic Literature Review. — Text: electronic // British Council: [site]. — URL: https://www.britishcouncil.org/sites/default/files/ai_in_english_language_teaching_systematic_review.pdf (accessed: 29.10.2024).
  2. Han, Bing. (2019). Application of Artificial Intelligence in Autonomous English Learning among College Students. International Journal of Emerging Technologies in Learning (iJET). 14. 63. 10.3991/ijet.v14i06.10157.
  3. Göçen, Ahmet & Aydemir, Fatih. (2020). Artificial Intelligence in Education and Schools. Research on Education and Media. 12. 13–21. 10.2478/rem-2020–0003.
  4. Cantos, Karina & Varas Giler, Rosa & Magayanes, Israel. (2023). Artificial Intelligence In Language Teaching And Learning. Ciencia Latina Revista Científica Multidisciplinar. 7. 5629–5638. 10.37811/cl_rcm.v7i4.7368.
  5. Artificial Intelligence and Use of AI in Education: Deciding on the Future We Want. — Text: electronic // UNESCO: [site]. — URL: https://www.unesco.org/en/articles/use-ai-education-deciding-future-we-want (accessed: 29.10.2024).
  6. Crompton H., Edmett A., Ichaporia N., Burke D. AI and English language teaching: Affordances and challenges // British Journal of Educational Technology. 2024. P. 1–27. https://doi.org/10.1111/bjet.13460
  7. Pak L. E., Kryukova A. A. On the Possibilities of Using Programs with Artificial Intelligence in Teaching a Foreign Language // Territory of New Opportunities. 2024. No. 2. URL: https://cyberleninka.ru/article/n/o-vozmozhnostyah-ispolzovaniya-programm-s-iskusstvennym-intellektom-v-obuchenii-inostrannomu-yazyku (date of access: 10/20/2024).
  8. Tolstykh, O. The potential of artificial intelligence in language teaching: practical recommendations for teachers. / O. Tolstykh. — Text: electronic // ResearchGate: [site]. — URL: https://www.researchgate.net/publication/372743869_Potencial_iskusstvennogo_intellekta_v_azykovom_obrazovanii_prakticeskie_rekomendacii_dla_prepodavatelej (date of access: 10/29/2024).
  9. Tenitsky, G. Application of artificial intelligence to improve education in secondary schools / G. Tenitsky. — Text: direct // Bulletin of Science. — 2024. — No. 4 (73).
Основные термины (генерируются автоматически): URL, TOEFL, UNESCO, CIS, ELT, IELTS, LLM.


Похожие статьи

Задать вопрос