Tajikistan at the crossroads of R&D: how zehn.ai is shaping the future of AI
Najima Najima Noyoftova, Director of A7σ Executive Office, spoke about the development of artificial intelligence in Tajikistan and how zehn.ai, a cutting-edge R&D lab that has the potential to become Tajikistan's first national patent in the field of artificial intelligence on the global stage, is contributing to it.
Tajikistan has a chance to become a vanguard
On 30 November 2022, the world changed. Since then, processing 2 billion queries every month, ChatGPT algorithms have made getting accurate analytical information instantaneous. And now, thanks to the adoption of artificial intelligence (AI) algorithms, companies are automating processes and reducing operational costs by a third, while employee productivity is doubling.
Thanks to AI, in the educational space, more than 75 per cent of students have found their way to knowledge that was previously out of their reach. And in this global race, Tajikistan has a real chance to become a vanguard. How?
At the beginning of 2023, I joined the team of zypl.ai, then a young fintech startup with a focus on credit scoring. At that time, zypl was just starting its journey to enter the GCC markets: first presentations at a banking conference in Dubai and first meetings with potential partners.
Recognising the speed of growth of AI algorithms, in parallel, zypl has launched its first R&D (research and development) project in partnership with foreign researchers.
The problem for our banking clients was that credit risk models based on historical data face serious limitations. These models predict the future based solely on the past, making them vulnerable to new, volatile circumstances.
The first zGAN concepts were developed back in 2021 at Stanford University by Azizjon Azimi, founder of zypl and chairman of the Artificial Intelligence Council under the Ministry of Industry and New Technologies of Tajikistan. Back then, at the height of the COVID-19 pandemic, new AI models often failed in real-world applications. The problem was not the amount of data available, but their lack of adaptation to global contingencies.
Subsequently, in creating the idea of zehn.ai, we realised that not only historical data, but also innovative approaches to creating data with modified distributions were needed to create truly robust models. This initiative is not just revolutionary; it has the potential to become Tajikistan's first national patent in artificial intelligence on the global stage.
There are outstanding minds behind the development of zehn.ai
Obtaining the first patent in the field of AI could be a turning point for Tajikistan, positioning our country as a frontrunner in the world of AI. We are already the first country in Central Asia to adopt the National Strategy for the Development of Artificial Intelligence until 2040, thanks to the initiative of the head of state Emomali Rahmon. Our leap into the world of AI is particularly important in the context that two-thirds of the world's countries have not even developed their National AI Strategies yet.
Stanford University, where the zGAN model originated, already recognises our AI initiatives. Last year, Stanford published a ‘case study’ research paper written by researchers at Stanford that details the history of the AI industry in Tajikistan through the lens of several key organisations.
In February this year, Azizjon Azimi and zypl.ai's chairman, renowned American investor David Halpert (founder of a multi-billion dollar hedge fund), visited Stanford as part of a case study presentation to students and professors.
Behind the development of zehn.ai are outstanding minds with high expectations centred on them. The lead developer of zGAN is Sergey Shulgin, a Russian expert and PhD candidate in economics who emigrated to Tajikistan in 2022. As part of zGAN development in our Dushanbe AI lab, he created a breakthrough prototype of a synthetic data model with outliers, which has already been successfully applied by zypl.ai's banking partners.
Our compatriot Shukhrat Khalilbekov, former lead AI engineer at companies such as Okko and Citymobile, and now VP of product at zypl.ai, is leading the project, providing coordination and strategic direction to the team.
Also involved in the development of zGAN is a local expert, Doctor of Science and Deputy Chairman of the AI Council, Akhliddin Nizomiddinov. The process of recruiting foreign specialists among Russian expats working in the field of AI in our laboratories in Dushanbe has already been launched.
Ilyas Varshavsky, a former researcher at the National Centre for Cognitive Development and Research Centre for Artificial Intelligence ‘Strong Artificial Intelligence in Industry’ of ITMO University, Tajikistan, PhD in Engineering, has recently joined the team. Ilyas joined our team after winning the hackathon ‘Disruptive DataHack’ organised by the AI Council.
From agriculture to e-government
zehn.ai is just beginning its journey. To date, zGAN has been implemented to develop macro-resilient credit risk models in a number of zypl's global partner banks: CredibleX (UAE), Commercial Bank International (UAE), Home Credit Bank (Kazakhstan), and Bank CenterCredit (Kazakhstan). By the end of this year, zehn.ai plans to finalise the architecture of zGAN applications in different areas of the economy, from agriculture to e-gov. Negotiations are already underway with leading technology companies from Latin America to the Middle East, as well as with government agencies in CIS countries for the application of zehn products.
zehn.ai's activities are in line with the vision of our National AI Strategy to achieve 5% of GDP from artificial intelligence by 2040. Exporting AI software from zehn.ai could dramatically change the trade balance structure of our economy. Stanford University estimates that more than 50 significant new AI models are created each year worldwide (AI Index Report).
Starting next year, by attracting top global and local AI researchers, we at zehn.ai plan to develop 3-5 new AI algorithm architectures and models each year (i.e. up to 10% of global volume), starting with zGAN. This is an ambitious target, which we see as absolutely achievable in the context of successful precedents of zGAN creation and established export chains of AI products from zypl.ai.
Just the other day, I witnessed how Tajikistan has already started exporting its AI solutions. The republic's Ministry of Industry and New Technologies and the government of the Emirate of Ras Al Khaimah (UAE) signed a memorandum under which the Emirate will develop its National AI Strategy based on Tajikistan's strategy.
In return, the Emirati side will provide direct investment in our country's green economy, as well as preferential accessibility to its large-scale port for the transit of Tajik goods. I had the honour of presenting the vision of launching Area AI, a large-scale AI technology park to be formed in our country in the next few years, to the high-ranking guests.
Our country is on the threshold of the most important stage in the world of artificial intelligence, which has a historical parallel with the significant innovations of our ancestors. The very modern concept of ‘algorithm’ comes from the name of the great Tajik scientist al-Khorazmi, who defined the contours of the development of mathematical sciences for centuries to come.
I am confident that the development of AI ecosystem in our country, with the vanguard role of zehn.ai, will become a symbol of historical restoration of Tajikistan's role as an advanced civilisation for scientific and technological progress in the world.