The advent of ChatGPT and other generative artificial intelligence (GAI) tools in late 2022 has ushered in a new era of AI-driven content creation. GAI is defined as a class of AI models capable of generating synthetic textual and visual data that closely resembles real-world data (Bandi et al., 2023). This technology, unlike its analytical AI counterpart, possesses a service-oriented nature, enabling its widespread application across various domains. Research suggests that GAI can revolutionize education by personalizing learning experiences and providing near-instant feedback, enabling automated assessment, and fostering more interactive and engaging lessons (Onesi-Oziganun, 2024). At the government level, the importance of AI technology is enshrined in national strategies and programs, e.g., United Kingdom (UK Government, 2023), European Union (European Commission, 2022), Russia (Garant, 2023). 

The potential benefits and risks associated with GAI highlight the critical need for adaptations by educational organizations. However, there is a risk that schools will continue to operate in a manner that ignores this new reality, as education is often viewed as a conservative system (Fuller, 2020). Previous waves of digitalization in education have encountered resistance from teachers (Wohlfart & Wagner, 2023). The COVID-19 pandemic, while forcing educational institutions to rapidly adopt digital technologies (Koroleva & Andreeva, 2024), also created a stressful environment for educators (Gómez-Domínguez et al., 2022). This experience, while exposing teachers to digital tools, may have inadvertently fostered a delayed resistance towards new technologies, potentially hindering the future adoption of AI. Furthermore, effective strategies for integrating AI into education remain unclear (Chiu et al., 2023), underscoring the need for careful consideration and targeted approaches to ensure successful AI implementation.

Rogers' Diffusion of Innovations Theory, widely used in education research, outlines the stages individuals, including teachers, traverse in adopting new practices. However, contemporary research debates whether individuals consistently follow these stages linearly and what communication channels involved at each stage (Achuthan et al., 2020; Miranda et al., 2016; Botha et al., 2018; Agélii Genlott et al., 2023). Rogers (2003) proposes a five-stage model of individual innovation adoption, progressing from knowledge and information seeking, through persuasion and attitude formation, to a decision to adopt or reject, implementation, and finally confirmation, which leads to continued use or discontinuance. In addition, Rogers proposes considering (1) prior conditions, (2) perceived characteristics of the innovation, and (3) communication channels when describing an individual's adoption process. While Rogers posits these factors as universal across stages, recent research suggests a need for further refinement in aligning factors with specific stages. This calls for a more nuanced understanding of how these factors interact and influence adoption decisions at different points in the process (Achuthan et al., 2020; Arthars & Liu, 2020). 

The use of AI remains non-mandatory for teachers in Russia (AI in Russia, n.d.). Despite the lack of widespread adoption, some educators have already begun utilizing GAI in their professional activities. This presents an opportunity to investigate the factors influencing their decision to adopt this technology, as well as the process of decision-making and implementation within their teaching practices. Understanding the experiences of these early adopters is crucial for informing future strategies aimed at promoting the responsible and effective integration of GAI into education systems.

Consequently, this study addresses the following research questions:

• RQ1: Does the adoption of GAI by school teachers within their professional practice follow a staged and sequential process, aligning with the stages outlined in Rogers’ Diffusion of Innovations Theory?

• RQ2: What specific communication channels are utilized by school teachers during each stage of GAI adoption, and what factors are seen as being crucial at each stage of this process?

Data was collected from December 2023 to March 2024 through 17 in-depth, semi-structured interviews with Russian school teachers actively using GAI in their professional practice. Participants were primarily recruited from the Digital Teacher Map (2024), which allowed filtering teachers based on their use of digital technologies, including AI. Of the 48 teachers listed, 14 agreed to participate. Three additional participants were later found through referrals. The teachers' backgrounds were diverse, with a mix of humanities and STEM subjects, varying teaching experience, and a range of additional responsibilities. The study sample comprised participants from diverse educational settings, encompassing schools of varying sizes, including specialized institutions (gymnasiums and lyceums) and regular schools, located across different regions of Russia and operating in both large cities and smaller settlements. This diversity of educational settings allowed us to capture a wide range of socio-demographic contexts and investigate the organizational and individual factors that facilitate teachers' adoption of new technology. The research presented in this paper was conducted using data and resources from the 'Mirror Laboratories' project, a collaborative initiative between the Higher School of Economics (HSE University) and Tomsk State Pedagogical University (TSPU). Interviews were conducted online via Zoom, lasting approximately 40-50 minutes each, and continued until data saturation was achieved (Strauss & Corbin, 2001). All participants provided informed consent for audio recording, data processing, and anonymous use of the data. A previously piloted guide was used, covering introductory information, GAI understanding and personal use, teacher activity during GAI adoption, unsuccessful practices, school context, and future visions of GAI in education. Data analysis was divided into two sub-stages: primary analysis (January-March 2024), concurrent with data collection, and subsequent analysis (May-June 2024). Primary analysis used open coding to establish an overarching vision for further coding (Denzin, 1989). The subsequent analysis followed multi-grounded theory principles with three stages: inductive coding, conceptual refinement, and pattern coding. Inductive coding, similar to open coding, involved analyzing data without pre-existing theoretical frameworks (Goldkuhl & Cronholm, 2010). Conceptual refinement then critically reflected on the inductive codes, identifying those aligned with Rogers' Diffusion of Innovations Theory. Pattern coding, akin to axial coding, established the stages of the innovation-decision process as a central "axis", with other codes developed around it.

The adoption of GAI by school teachers within their professional practice follows a staged and sequential process, aligning with the stages outlined in Rogers’ Diffusion of Innovations Theory. The analysis shows that this process can be more complex at the confirmation stage, comprising three substages: re-persuasion, re-evaluation of decision and true confirmation. Additionally, different groups of informants can be identified based on the end-points in their innovation-decision process, and these end-points happen to be at persuasion and re-persuasion stages of the process. The research delves deeper into the confirmation stage by identifying three substages: re-persuasion, re-evaluation of decision, and true confirmation, highlighting the critical role of re-persuasion. In general, it can be said that both stages of persuasion and re-persuasion, which are characterized by the formation of attitudes, are crucial, as they often represent points of potential drop-off in the innovation-decision process. This signifies the paramount importance of attitudes in shaping teacher adoption decisions, aligning with findings from other studies (Holdsworth & Maynes, 2017; Ajzen, 1991), which emphasize the significant role of teachers’ attitudes towards innovations in their adoption, including AI adoption (Crompton et al., 2022; Chiu et al., 2023; Su et al., 2023). The study further clarifies the specific communication channels and factors influencing the innovation-decision process at each stage with a focus on school teachers. Although factors are described widely in various studies (Al Darayseh, 2023; Ayanwale et al., 2022), the novelty of this work was the description of factors at each stage as different factors contribute to the process at different time. The research offers valuable insights for school leaders. It pinpoints the stages at which school teachers require additional support and highlights the most effective communication channels and the most crucial factors for providing this support.

Achuthan, K., Nedungadi, P., Kolil, V., Diwakar, S., & Raman, R. (2020). Innovation adoption and diffusion of virtual laboratories. International Journal of Online and Biomedical Engineering, 16(09), 4–25. https://doi.org/10.3991/ijoe.v16i09.11685 Agélii Genlott, A., Grönlund, Å., Viberg, O., & Andersson, A. (2023). Leading dissemination of digital, science-based innovation in school–a case study. Interactive Learning Environments, 31(7), 4171–4181. https://doi.org/10.1080/10494820.2021.1955272 Al Darayseh, A. (2023). Acceptance of artificial intelligence in teaching science: Science teachers' perspective. Computers and Education: Artificial Intelligence, 4, Article 100132. https://doi.org/10.1016/j.caeai.2023.100132 Bandi, A., Adapa, P., & Kuchi, Y. (2023). The power of generative AI: A review of requirements, models, input–output formats, evaluation metrics, and challenges. Future Internet, 15(8), 260. Chiu, T. K., Xia, Q., Zhou, X., Chai, C. S., & Cheng, M. (2023). Systematic literature review on opportunities, challenges, and future research recommendations of artificial intelligence in education. Computers and Education: Artificial Intelligence, 4, Article 100118. https://doi.org/10.1016/j.caeai.2022.100118 Crompton, H., & Burke, D. (2022). Artificial intelligence in K-12 education. SN Social Sciences, 2(7), Article Holdsworth, S., & Maynes, N. (2017). “But what if I fail?” a meta-synthetic study of the conditions supporting teacher innovation. Canadian Journal of Education/Revue canadienne de l'éducation, 40(4), 665-703.https://www.jstor.org/stable/90018384 Miranda, M. Q., Farias, J. S., de Araújo Schwartz, C., & de Almeida, J. P. L. (2016). Technology adoption in difusion of innovations perspective: introduction of an ERP system in a non-profit organization. RAI Revista de Administração e Inovação, 13(1), 48–57. https://doi.org/10.1016/j.rai.2016.02.002 Onesi-Ozigagun, O., Ololade, Y.J., Eyo-Udo, N.L., & Ogundipe, D.O. (2024). Revolutionizing education through AI: a comprehensive review of enhancing learning experiences. International Journal of Applied Research in Social Sciences, 6(4), 589–607. https://doi.org/10.51594/ijarss.v6i4.1011 Rogers, E.M. (2003). Diffusion of Innovations (5th ed.). Free Press. Su, J., Guo, K., Chen, X., & Chu, S. K. W. (2023). Teaching artificial intelligence in K–12 classrooms: a scoping review. Interactive Learning Environments, 1–20. https://doi.org/10.1080/10494820.2023.2212706 Wohlfart, O., & Wagner, I. (2023). Teachers’ role in digitalizing education: an umbrella review. Educational Technology Research and Development, 71(2), 339–365. https://doi.org/10.1007/s11423-022-10166-0