AIED IN HIGHER EDUCATION AND THE ARMENIAN ACADEMY: A LITERATURE REVIEW

Artificial intelligence represents technology that mimics human intelligence in a manner that employs machine perception and learning as well as evolutionary algorithms to accomplish predefined goals. AIEd specifically concerns the applications that assist in (re-)processing of information for comprehension by students, instructors, and administrators during a course’s delivery, administration, and learning. Consequently, it is important to briefly mention four general AIEd applications identified and studied in prior research. They involve (a) student profiling and performance prediction; (b) assessment, evaluation, and performance flagging; (c) spontaneous programming adaptations; and (d) smart tutoring systems (Crompton and Burke Reference Crompton and Burke2023; Zawacki-Richter et al. Reference Zawacki-Richter, Marín, Bond and Gouverneur2019). Research generally indicates that although AIEd does not itself improve direct student learning outcomes, it instead serves a labor-saving role for faculty in administrative functions and highlights downward student performance for intervention and correction by instructors (Bates et al. Reference Bates, Cobo, Mariño and Wheeler2020, 2).

But these applications have grown more prominent in advanced economies only recently and represent the most frequently used and researched applications of AIEd in advanced economies (Crompton and Burke Reference Crompton and Burke2023; Zawacki-Richter et al. Reference Zawacki-Richter, Marín, Bond and Gouverneur2019). Initially, AIEd was used almost exclusively by computer science departments, STEM fields, and by university administrations, whereas all others deployed it at peripheral levels. Teachers’ use of AIEd has grown only fairly recently (Hinojo-Lucena et al. Reference Hinojo-Lucena, Aznar-Díaz, Cáceres-Reche and Romero-Rodríguez2019; Zawacki-Richter et al. Reference Zawacki-Richter, Marín, Bond and Gouverneur2019, 7–10). Prior to contemporary advances, teacher-facing and student-facing AIEd alike remained relatively underdeveloped, possibly owing to their primary development and use by computer scientists (Akgun and Greenhow Reference Akgun and Greenhow2022; Bearman, Ryan, and Ajjawi Reference Bearman, Ryan and Ajjawi2023). Specifically, AIEd in higher education today primarily serves teaching faculty during course preparation and administration, but it is often ignored or unused in class delivery or for collection of live student feedback (Bearman, Ryan, and Ajjawi Reference Bearman, Ryan and Ajjawi2023; Celik et al. Reference Celik, Dindar, Muukkonen and Järvelä2022). As AIEd is still gaining momentum and footholds within academia, its widescale use is anticipated in the future (Celik et al. Reference Celik, Dindar, Muukkonen and Järvelä2022).

Faculty use cases must also be balanced and prioritized with students’ access to AIEd, but students also need to be educated carefully on its capabilities and limitations. Directing students to productive and enriching uses of AI ought to be encouraged as inoculation against its use as a crutch and/or replacement for their own studies (Jarrah, Wardat, and Fidalgo Reference Jarrah, Wardat and Fidalgo2023; Popenici and Kerr Reference Popenici and Kerr2017; Sullivan, Kelly, and Mclaughlan Reference Sullivan, Kelly and Mclaughlan2023). In this vein, popularly, misuse of and poor education in the use of AIEd can decrease student and academic integrity and increase misconduct during completion of routine assignments and work and thus can threaten long-term outcomes. (Jarrah, Wardat, and Fidalgo Reference Jarrah, Wardat and Fidalgo2023; Sullivan, Kelly, and Mclaughlan Reference Sullivan, Kelly and Mclaughlan2023). However, efficiency gains for instructors and students contain risks of their own besides individual use.

When implemented and scaled within school districts across entire regions, AIEds’ mass algorithmic power risks homogenizing and abandoning students by applying solutions tailored to enriched and powerful social strata that are overidentified in models’ training data as a society’s “average” students (Holmes and Porayska-Pomsta Reference Holmes and Porayska-Pomsta2022). This sampling error can yield pernicious, discriminatory outcomes that exacerbate and reflect real structural inequalities in the social world. However, such issues often go unstated within AIEd models, as socially marginalized groups—especially racial and gender groups—are blocked from proper representation in the samples used during AIEd training. Still worse, AIs and AIEds are not static, and the varied use of other programs—such as facial and vocal recognition tools—can accentuate marginalization of discriminated groups and can perpetuate structural racial and gender biases (Benjamin Reference Benjamin2019; Buolamwini Reference Buolamwini2023). These issues can further magnify existing biases in educational systems that already overwhelm youth, women, racialized, and minoritized student influence and stakes within such systems (Brossi, Castillo, and Cortesi Reference Brossi, Castillo, Cortesi, Holmes and Pomsta-Porayska2023; Treviranus Reference Treviranus, Holmes and Pomsta-Porayska2023).

Although AIEds can work as “springboards” that could advance the (most catered) students, impoverished districts, populations, and regions could implement poorer quality and lower-tiered AIEd models that compromise the application of their expected uses in the classroom via poorly fitted implementation, weaker pedagogical or programming suggestions, decreased privacy, and weaker nudging toward established expectations (Bartoletti Reference Bartoletti, Holmes and Pomsta-Porayska2023). Advancements in AIEd by educational-technology corporations paired with governments seeking cost-cutting “solutions” potentially lead to disassembly, disempowerment, and elimination of teachers themselves, thereby corroding and potentially destroying critical nodes for communities’ transfer of knowledge, skills, socialization, and socioeconomic transformation (Braverman Reference Braverman1974; Noble Reference Noble1995, Reference Noble2003).

Within Armenia, Soviet legacies and post-Soviet neoliberalization have affected the administrative and financial circumstances of the country’s higher education institutions. Privatizations and mass withdrawal of state funding, alongside economic difficulty, have led state universities to endure large funding cuts amidst the establishment of private institutions and foreign branch campuses. By the late 2000s, total state funding accounted for an average of only 23% for all higher education institutions’ budgets, with student tuition providing the remainder. At private institutions, student tuition filled 94%–100% of university budgets (Karakhanyan Reference Karakhanyan, Huisman, Smolentseva and Froumin2018, 61–62). Currently, 71,732 students study in Armenian universities, of which 59,169 hail from public universities and 12,563 reside in private universities (Statistical Committee of the Republic of Armenia 2023).

Nonetheless, many changes have transpired to drive the Armenian academy toward improved outcomes, quality, and international recognition. Many domestic reforms permitted interfacing with international institutions, linkage with Erasmus, engagement in international exchange and recruitment programs, and attention to research and innovation. There is also an increasing focus on the introduction of technology into higher education, including e-learning platforms, online courses, and other digital tools to improve the quality of education. This includes standardizing qualifications, degree structures, and credit systems under the rubrics of the European Credit Transfer and Accumulation System as well as quality assurance to facilitate the mobility of students and academic staff (Aperyan Reference Aperyan2021).

The geographical and social position of Armenia also favor multilingual and multicultural education. Although Armenian is the main language of instruction in most universities in the country, the importance of knowing English and other foreign languages is noteworthy. The American University of Armenia notably delivers instruction in English, whereas the French University in Armenia and Russian-Armenian University deliver instruction in French and Russian. To attract international students, many universities, such as Yerevan State University, Armenian State University of Economics, the Mkhitar Heratsi Yerevan State Medical University, and others also offer various educational programs in English. Efforts toward educational reforms have also been made by the Government of Armenia (Dobbins and Khachatryan Reference Dobbins and Khachatryan2015). Current challenges facing higher education in Armenia include the status of public financing (Karakhanyan Reference Karakhanyan, Huisman, Smolentseva and Froumin2018), adequate infrastructure, and improved cooperation between the academy and employers (K. Atoyan, Babajanyan, and Atoyan Reference Atoyan, Babajanyan and Atoyan2021; Keryan et al. Reference Keryan, Muhar, Mitrofanenko, Khoetsyan and Radinger-Peer2020). In addition, brain drain has also been a constant concern in recent decades, as many highly qualified professionals, including those from the academic field, leave the country in search of better opportunities abroad (V. Atoyan Reference Atoyan2021).

DATA AND METHODS

During 2021–2022, 17% of AI research in higher education globally focused on academia (Crompton and Burke Reference Crompton and Burke2023, 1). To contribute to this literature, the research group conducted a sociological survey during August–September 2023 among the social sciences educators of universities in Armenia. A comprehensive list of universities whose representatives have responded to the request is available in the appendix. The decision to focus on social sciences was driven by the unique role these disciplines play in understanding and shaping educational practices and policies within the country, including AIEd use. Further surveys among educators in other disciplines remain to be completed.

The survey was initially designed as a Google Form and emailed to university faculty for ease of access. It included questions designed to gauge faculty perceptions, use, and concerns regarding AIEd. The survey was disseminated via email to 892 faculty at various universities throughout Armenia. To increase participation across different age groups and ensure a comprehensive data set, printed versions of the survey were also distributed. The survey was intended to capture responses from at all universities in Armenia regardless of their status. Participants included faculty with graduate degrees, both with and without doctorates.

The selection of the sample was purposive, targeting educators in social sciences within higher education institutions across Armenia. A total of 184 faculty (20.6% response rate) from eleven higher education institutions participated, representing a wide section of Armenian higher education, where the majority of postsecondary Armenian students study (81.5%). Responses were gathered from faculty at nearly all of the highest-ranked universities in Armenia. According to the latest national ranking, the leading universities in Armenia are Yerevan State University, Armenian-Russian University, and the M. Heratsi Yerevan State Medical University (NCET 2016). Other ranking criteria point to the Armenian State University of Economics and the Armenian National Agrarian University, among others (Yerevan State University 2017).

Although the survey was completed mainly by educators at state universities, it should be noted that educators at state universities also very often teach in nonstate universities. The ratio of the number of faculty of in Armenian universities is as follows: 9,574 educators teach in state universities and 1395 in nonstate universities (Statistical Committee of the Republic of Armenia 2023). The faculty who participated in the survey indicated their primary affiliation. Responses were excluded if they were incomplete or if the participants did not meet the criteria of being current social sciences educators in Armenian higher education institutions. Additionally, any duplicate responses identified through data cleaning were removed to ensure the accuracy and reliability of the data.

The survey responses were collected and analyzed to provide insights into the state of AI adoption in Armenian higher education. The demographic information of the respondents, including age and scientific degrees, is summarized in figures 1 and 2. It should be noted for clarification that the titles “Candidate of Sciences” and “Doctor of Sciences” are postgraduate degrees used in many post-Soviet countries. Although they are each similar to the PhD and higher doctorate degrees used in Western countries, there are differences in their structure and significance. The Candidate of Sciences is an advanced research degree, similar to a PhD; whereas, the Doctor of Sciences is a higher doctorate awarded for exceptional scholarly achievements.

Figure 1

Respondents’ ages

Figure 2.

Scientific degrees of the respondents

This study complies with the APSA Principles and Guidance for Human Subject Research. All participants were informed about the purpose of the research, and their consent was obtained before they participated in the survey. Confidentiality and anonymity of the respondents were strictly maintained throughout the study.

It is important to clarify that the age distribution of respondents, shown in figure 1, does not directly indicate the prevalence of AIEd use among different age groups. The age data were collected to provide demographic context, and this information is not meant to be interpreted as an indicator of AIEd adoption by age. The survey did not specifically cross-tabulate age with the usage of AIEd technologies. Thus, any analysis on the correlation between the age of educators and their use of AIEd would require a more detailed statistical examination beyond the scope of this study.

FINDINGS

Following the completion of the survey, which was designed to capture the attitudes of Armenian teaching faculty toward AIEd and their extent use of the technology, the following results were captured.

1. Faculty Concerns about AI

As shown in table 1, rough parities among teaching faculty regard AIEd as either not a threat or as a threat to their futures in academia, with clear pluralities emerging only where existential outcomes—such as the fate of human civilization (50%) or specter of mass unemployment (46.7%)—were concerned. In other questions, the major scale and perceived potential to transform society with its automating capabilities of AIEd is regarded evenly as a threat or no threat to employment across the entire labor market, to academia, and to university faculty worldwide, as well as in Armenia.

Table 1.

Faculty Concerns about AI

2. Survey Findings: Faculty Use of AI in Courses

Table 2 shows respondents’ use of AIEd for teaching. In practice, 58.7% of respondents use AIEd in their courses, but only to a “small extent” in mixed use cases. Course assessment, planning, and administration all emerged as prominent uses of AIEd, with only 6.5% of respondents noting its direct use for educational applications. Roughly 40% eschewed use of the technologies in their courses.

Table 2.

Faculty Use of AI in Courses

3. Survey Findings: Faculty Views toward AI Use

Although participants expressed worry and lack of concern at roughly equal levels in prior questioning, attitudes are generally favorable regarding the potential of AIEd use in Armenian classrooms to somewhat improve or aid their students’ learning outcomes (table 3). Independent thinking and research skills, coherency of educational material, objective student assessment, facilitation of learning, advancement of national higher education, and moderately improved student-teacher dialogues all receive majorities of respondents that regard AIEd as capable of bringing partial potential improvements.

Table 3

Faculty Views toward AI Use