The Armenian Mirror-Spectator. WATERTOWN — It seems like all aspects of artificial intelligence (AI) are simultaneously fascinating and frightening the world as its role rapidly grows in our daily lives. Armenia is no exception, and there FAST (Foundation for Armenian Science and Technology) is attempting to create a new generation of innovators and spur the development of this important field. Its first step is through what it calls the Generation AI High School Project, developed and implemented with the Armenian Ministry of Education, Science, Culture and Sports (henceforth, Ministry of Education).
Suzanna Shamakhyan
FAST’s executive director, Suzanna Shamakhyan, during a visit to Boston this summer, explained that this three-year merit-based program was launched in 2023 in 15 schools, and in 2024, 9 of those 15 schools had a second cohort enroll. This September the program is in 23 schools throughout Armenia, with around 900 students in all three high school grades in parallel. Next year it will spread to up to 45 schools throughout the country. By then, any high schooler who wants to choose the program could find a high school offering it.
The program is demanding, with advanced placement level math, statistics and deep machine learning, and students must take qualifying exams each year to be allowed to continue. In the upcoming spring, around 180 will be the first graduates of the program. Consequently, FAST is preparing a program at the undergraduate level to allow these graduates a continuity in their education.
Shamakhyan pointed out that generally, other programs in the field are not integral parts of the school curriculum but are afterschool programs. For example, STEP.ai is organized by the Synopsys Foundation, AGBU, and the Union of Employers of Information and Communication Technologies (UEICT) and is accredited by the Ministry of Education. Its focus is teaching applications of AI and is primarily done online with videos and teacher trainers. The Children of Armenia Fund (COAF) AI offerings in its Smart Centers include Geographic Information Systems (GIS) courses with some AI topics. COAF and FAST are discussing how their students can collaborate on joint projects.
Read also
FAST’s approach is to provide the foundation of math and computer science to teach how machine learning works and algorithms are created. Shamakhyan said that the various programs could be considered complementary and do not compete with each other.
FAST’s goal is to make Armenia a STEM or AI hub, and Shamakhyan said that FAST realized through its prior efforts that the way to do this was through changing Armenia’s educational system. Before its Generation AI program, it did a lot of afterschool programs for university students. Shamakhyan said the FAST people realized that even after doing such programs for years, it could not change the overall system in this manner. She continued, “When we were trying to understand what the root cause of the problem is — why we don’t have the talent pipeline at scale — we ended up understanding that the school is one of the main bottlenecks.”
Generation AI high school students with FAST co-founder and board member Dr. Noubar Afeyan
A handful of expensive private schools in Yerevan provide specialized math education. To get a larger stream of students into this field who eventually will either become AI innovators or part of a highly capable workforce for industry, FAST approached Armenia’s Ministry of Education.
Shamakhyan said that the ministry had just approved a strategy for 2030 and was attempting major reforms with the support of the World Bank and European Union in STEM education, but had some challenges in implementation because its system did not have the capacity yet to execute these reforms on this scale. FAST therefore structured its AI program to connect with the larger reforms, which would make it easier for the government to commit to it and take ownership. The World Bank will be funding the creation of all the AI labs that will be needed for the Generation AI program’s schools, estimated to cost about $1.2 million, she said.
FAST’s team is very aware of the issues of working with a bureaucracy, Shamakhyan said, and with both the government and FAST in agreement on objectives, she opined that “we have a strong public-private partnership,” with the ministry considering Generation AI as its own program. There are over 40 partners, and around 80 experts contributing to the content, development, methodology and teaching of this program, both from the diaspora and locally. She added, “It also has been an interesting way to demonstrate how you can actually translate the diaspora’s knowledge into actual system reform.”
The United Nations Educational, Social and Cultural Organization (UNESCO) meanwhile published a framework for AI creators’ education, which is basically what FAST does, and invited FAST in 2024 to Paris to showcase Armenia as a case study on how to build curricula. From then on, Shamakhyan said that the Armenian government and FAST have maintained close ties with UNESCO. Armenia’s example has now been presented at four UNESCO conferences (most recently in Paris), and other international organizations are interested in its statistics.
The Organization for Economic Cooperation and Development (OECD) recently published its own AI literacy framework and FAST is in discussions with it and the Armenian Education Ministry to see how AI literacy content (as opposed to AI creators’ education) can also be embedded into the country’s school system to prepare the general populace for living in an AI world.
Shamakhyan said, “I can tell you from what we are observing — because we participate constantly in AI education conferences and go to places like that — that we are in the forefront of all this.”
When FAST began to work on the curricula in 2022 and 2023, only China and one district with a pilot program in the state of Georgia in the US, were doing AI creators’ education and AI literacy programs.
She noted that globally, China, India and some Scandinavian nations are aggressively pursuing AI education. The rest of the world at present is only focusing on AI literacy.
In other words, Armenia, particularly as a small country, stands out for its ambitious AI creators’ education plans, along with the priority the government, nonprofits like FAST, and the broader ecosystem place on ensuring that AI literacy levels increase. “We have faster conversion because Armenia is small. We have greater chances to do this more affordably and faster,” she said.
Shamakhyan also hopes that after Armenia is made AI-ready, FAST will be able to begin AI literacy courses in the diaspora to make all Armenian people educated in AI.
In Armenian public schools, there is a modular-based education system with tracks or specializations. The Generation AI program is for those students with a physics and mathematics high school major or specialization. Usually geometry, algebra (including calculus and statistics in different grades) and physics classes are mandatory, with a disproportionately high number of hours.
In the FAST program, regular algebra, around 5-6 hours a week, is replaced by what Shamakhyan calls a very intensive contemporary methodology of active learning in mathematics and project-based learning. This provides a good foundation for AI since basically AI is math and statistics, and as Shamakhyan pointed out, “Whatever buzz happens at the moment, we keep math as a core knowledge that will allow these kids become whatever they want to become regardless of what disruption takes place next time.”
Besides the revamped math classes, computer science and machine learning are added as separate subjects in extra hours. Shamakhyan said that each year of the three-year high school curriculum has around 350-400 hours for math, computer science, and AI (machine learning or deep learning), which means about half the students’ week is spent in these classes.
In addition, because the Ministry of Education has introduced project-based learning as a mandatory component of high schoolers’ learning, teachers have the opportunity to do this in after school classes, one more hour per day after the usual 9-2 or 9-3 schedule.
Preparation and Support of Teachers
In order for Generation AI to work, the teachers themselves have to be trained first, Shamakhyan said. Math teachers receive training in statistics, theory of probability and other new items in the curricula. Then they receive pedagogical training, as the old Soviet math teaching methodology, while it was at a high level, is outdated in many ways. Instead of assignment work, Shamakhyan said, the new approach includes active learning and real life mathematics.
There is also a teacher supervision program in which in addition to training in methodology and content, an external experienced teacher is hired to help the teachers do their weekly work. This is not mentorship, Shamakhyan said, but supervision, during which the external teacher sits down to help the local teacher do weekly planning and structure the administrative work to lighten the load. This support program for teachers lasts for three years.
She said that some of the math teachers, especially in the provinces, have been neglected for decades, and although this program does not pay them anything financially, “because we give them this support and materials, because they do something new and see that students have progress, they are so motivated. It is pretty inspiring. When you are a teacher by calling, just a little bit of support, acknowledgment and praise, and changing the perception of the teacher’s role in the community, has a big, big impact on their performance.”
There are no AI teachers in the schools. Shamakhyan explained that they are recruited from industry and university faculties, so they usually are young professionals and professors or recent doctorate holders, who teach AI around four hours a week. FAST determined that it was more efficient to bring in people with the necessary knowledge and train them in pedagogy. Afterwards they are connected with a school computer science teacher so that the teacher does the classroom control and pedagogy, while the outside professional provides more of the content.
The computer science teachers in turn receive help like the math teachers but receive much more knowledge training. As part of the Education Ministry’s STEM reform program, computer science is being embedded in courses starting as early as second grade and continuing all the way till the end of high school, but there are not enough teachers yet so FAST was asked to structure a national teacher training strategy for them.
In connection with educational reforms in general, Shamakhyan said, “I think there is no way Armenia can do it without the support of diasporan-led entities and the private sector…because the institutional strength is not there yet.”
The Ministry of Education is doing large-scale reforms in many directions and the system needs a lot of support both in terms of expertise and implementation, she said. FAST, she stressed, “is an exception as an entity that prefers to work with the system because, at the moment, when you look at what is going on, most of the programs … are outside of the system because everyone is saying, ‘I don’t want to deal with this.’” The latter approach will just lead to a lot of silos but not greater general change, she remarked. At the same time, she highlighted that many are willing to partner or provide support if they don’t have to directly bear the burden of working with the bureaucracy.
Shamakhyan is optimistic overall, declaring: “I think you can have remarkable results in a very short period of time because we don’t need thousands of teachers. Armenia is tiny. … Second, I think if we rethink a little bit about how we do the teaching and what the teachers’ role is, that can also change a lot of things, because they don’t need to be the smartest person in that field… because there is so much content that already exists… You can overcome some of these things relatively cheaper than in many other bigger established systems.”
In fact, in Armenia, she said, “It actually worked faster than we were anticipating. We were not planning to scale anything before 2026. The idea was that we would finish one cycle in 2026 and only after that would we scale [up], but because we see a momentum in the system, people don’t resist actually.”
Higher Education and Industry
Only a small portion of the high school graduates will be able to study abroad at top universities, so in order to ensure a continuation of the learning journey of the high school students, FAST has just created a program committee of diasporan Armenians and non-Armenians to prepare an undergraduate program in Armenia. This will be launched in 2026, Shamakhyan revealed.
With educational reforms being promulgated in Armenia, a new law will be published soon concerning higher education. This may allow teaching in English and using a lot of visiting professorship models and joint degrees. After the undergraduate program, FAST will focus on a doctoral one so it will be ready for those who will graduate the undergraduate one. Shamakhyan said it will only then do a master’s program, because there already exist some of these programs now.
The model FAST used with its Advance Research Grants program supporting a diasporan researcher as principal investigator in a project with a research team in Armenia nurturing junior researchers will be a useful framework for international research and put it at a high level. Armenia does not have enough AI senior researchers right now and therefore there are not sufficient doctoral thesis supervisors.
A current problem of the tech sector in Armenia, Shamakhyan said, is that senior specialists are needed but the short-term needs of industry do not allow them to develop since industry headhunts them at an early age. She said that since Armenia can never compete in quantity, it needs to focus on quality. This requires allowing specialists enough time and opportunity to become high-end experts. FAST, she indicated, is currently in talks with a number of big tech companies on how they could support the undergraduate programs in this sense.
The three ideal scenarios are students ending up in a doctoral program and working in a research center; continuing advanced education up to either a master’s or doctoral level and then conducting industrial R&D for big tech companies; or becoming entrepreneurs founding startup companies, which go through existing incubation programs in Armenia.
The remaining students will just become members of a highly capable workforce for the industry. Shamakhyan said, “And yes, we hope that by saying that we have a sustainable and scalable talent pipeline, this will make it easier to attract businesses to Armenia.” This would solve the recruitment problem that companies at present have due to the lack of people to work in this sector.
Armenia would also become a center of AI education for non-Armenians, Shamakhyan said, adding, “I think we have the luxury of trying to do that because of the diaspora. We have amazing professors in this field globally. If we structure this in the right way, we will be bringing top professors to Armenia, which will make it attractive for the international student.” By 2027, if not earlier, she said it will be realistic to target such students.
Educational Alliances
In addition to its relationship with the Armenian government, FAST works closely with a group of Armenian organizations in the educational field. In the summer of 2024, it and six other diasporan-led well-established organizations began planning a coalition, which was publicly announced in spring 2025. The 405 Educational Alliance (named in honor of the date of creation of the Armenian alphabet) includes the Armenian Educational Foundation, Armenian General Benevolent Union Armenia Branch , the Armenian Missionary Association of America, Ayb Educational Foundation, FAST, Simonyan Educational Foundation’s TUMO Center for Creative Technologies and Teach for Armenia Educational Foundation.
405 Educational Alliance public launch event, with Suzanna Shamakhyan in the red jacket
Shamakhyan said that though they all were dedicated to Armenia and knew their roles concerning Armenia’s educational sector, they did not work together a lot, while uniting efforts could accomplish much more. She said, “We hope to also break the notion that Armenians don’t collaborate. I don’t believe in that.”
After creating this alliance, the members are trying to do the right diagnoses of the strengths and weaknesses of the educational system, and then in the next phase deploy new solutions. A membership framework will also be launched so others can join in. “The utopic objective,” she said, “is that we will be able to shape a new model of education for Armenians globally.”
The challenge of preserving the language outside of Armenia is one of the topics being discussed. She said, “The idea is, what if we bring the best of diasporan and Armenian brains together, pilot projects in Armenia – a sort of proof of concept because there is so much space for that kind of thing in Armenia – and then scale them globally. So it is bringing in, and taking out.”
“I think that in the next ten years if the coalition becomes what we hope for it to become it will be a big transformation for Armenia in the space of education, and education impacts everything else,” she concluded.