According to the association's research, the larges
Posted: Thu Jan 23, 2025 3:57 am
At a press conference on October 4, 2023, dedicated to the study of the FinTech Association (FTA) on the use of artificial intelligence (AI) in the financial market, the association's CEO Maxim Grigoriev said that large language models (LLM) became the driver of development in 2023 in fintech, however, since their creation requires significant resources and access to a large amount of data, only a few organizations in Russia have them: YandexGPT at Yandex, FRED-T5, ruGPT-3 and NeONKA at Sber.
"As a result, 82% of respondents bolivia whatsapp resource are concerned about the shift of power towards large players and are interested in creating equidistant industry-wide large language models," noted Maxim Grigoriev.
t Russian banks by assets at the end of 2022 - Sberbank, VTB, Gazprombank, Alfa-Bank, Rosselkhozbank - invested a total of about $1 billion per year in the development of all AI solutions, which is about 80 billion rubles as of July 2023, while medium and small Russian financial companies invested an average of 100-300 million rubles per year in the implementation of a portfolio of projects with artificial intelligence.
According to Maxim Grigoriev, in the future, companies that can afford to develop LLM will gain a huge competitive advantage, and this inequality will grow. "The creation of an equidistant specialized language model for the financial sector, for example, by a consortium of financial organizations and technology companies, will help maintain competition and ensure the development of the financial market," said the head of the FinTech Association.
Grigory Gryaznov, head of the Analytical Services division of DOM.RF, noted that small companies do not have the opportunity to invest in the creation of large language models, since this requires expensive data processing centers (DPCs). According to him, in order to implement LLM in processes, fintech companies must first come to terms with the oligopoly and learn to integrate such models as SaaS services.
"As part of a trilateral agreement on the platform of the FinTech Association with Yandex, we are implementing a pilot project to introduce YandexGPT to solve internal problems," said Grigory Gryaznov.
Read also
GPT model developers have found business models for them
Yandex.Cloud LLC has identified six monetization scenarios for Yandex GPT, the Russian analogue of Chat GPT. Yandex Cloud has summed up the results of closed testing, in which 800 companies participated. The company received the most requests for the implementation of the language model in business processes from the IT, retail and banking sectors.
An analysis of practical cases of AI application in Russian fintech showed that 87% of the AI-based solutions used are aimed at data analysis and 63% of solutions are aimed at working with text: most often, financial companies use AI technologies for risk management and scoring, in recommendation systems in retail sales, as well as in decision support systems for customer service, including chatbots and knowledge bases in call centers.
According to the AFT, 84% of surveyed companies noted that they currently use decision trees, and 63% - convolutional neural networks. "We are seeing a trend towards reprioritization of the technologies and AI methods used in Russian fintech. By the end of 2023, 47% of study participants plan to implement generative pre-trained transformers, and 37% - graph neural networks," said Marianna Danilina, head of the research and analytics department at the FinTech Association.
A survey conducted by the AFT found that while the majority of fintech companies have implemented AI-based solutions, only 35% of companies surveyed have an approved AI strategy, with a further 12% including AI considerations in other organisational strategic documents.
"The key barrier to AI development in Russian fintech is the personnel shortage," the association's analysts emphasize. According to their data, 84% of surveyed companies faced a shortage of specialized AI specialists. The next most popular problems, according to the study, were long project implementation times (67%) and a lack of data for training models (61%).
The study covered 75% of the top 20 largest Russian banks: the FinTech Association conducted 45 in-depth interviews and studied more than 100 cases of AI application in fintech
"As a result, 82% of respondents bolivia whatsapp resource are concerned about the shift of power towards large players and are interested in creating equidistant industry-wide large language models," noted Maxim Grigoriev.
t Russian banks by assets at the end of 2022 - Sberbank, VTB, Gazprombank, Alfa-Bank, Rosselkhozbank - invested a total of about $1 billion per year in the development of all AI solutions, which is about 80 billion rubles as of July 2023, while medium and small Russian financial companies invested an average of 100-300 million rubles per year in the implementation of a portfolio of projects with artificial intelligence.
According to Maxim Grigoriev, in the future, companies that can afford to develop LLM will gain a huge competitive advantage, and this inequality will grow. "The creation of an equidistant specialized language model for the financial sector, for example, by a consortium of financial organizations and technology companies, will help maintain competition and ensure the development of the financial market," said the head of the FinTech Association.
Grigory Gryaznov, head of the Analytical Services division of DOM.RF, noted that small companies do not have the opportunity to invest in the creation of large language models, since this requires expensive data processing centers (DPCs). According to him, in order to implement LLM in processes, fintech companies must first come to terms with the oligopoly and learn to integrate such models as SaaS services.
"As part of a trilateral agreement on the platform of the FinTech Association with Yandex, we are implementing a pilot project to introduce YandexGPT to solve internal problems," said Grigory Gryaznov.
Read also
GPT model developers have found business models for them
Yandex.Cloud LLC has identified six monetization scenarios for Yandex GPT, the Russian analogue of Chat GPT. Yandex Cloud has summed up the results of closed testing, in which 800 companies participated. The company received the most requests for the implementation of the language model in business processes from the IT, retail and banking sectors.
An analysis of practical cases of AI application in Russian fintech showed that 87% of the AI-based solutions used are aimed at data analysis and 63% of solutions are aimed at working with text: most often, financial companies use AI technologies for risk management and scoring, in recommendation systems in retail sales, as well as in decision support systems for customer service, including chatbots and knowledge bases in call centers.
According to the AFT, 84% of surveyed companies noted that they currently use decision trees, and 63% - convolutional neural networks. "We are seeing a trend towards reprioritization of the technologies and AI methods used in Russian fintech. By the end of 2023, 47% of study participants plan to implement generative pre-trained transformers, and 37% - graph neural networks," said Marianna Danilina, head of the research and analytics department at the FinTech Association.
A survey conducted by the AFT found that while the majority of fintech companies have implemented AI-based solutions, only 35% of companies surveyed have an approved AI strategy, with a further 12% including AI considerations in other organisational strategic documents.
"The key barrier to AI development in Russian fintech is the personnel shortage," the association's analysts emphasize. According to their data, 84% of surveyed companies faced a shortage of specialized AI specialists. The next most popular problems, according to the study, were long project implementation times (67%) and a lack of data for training models (61%).
The study covered 75% of the top 20 largest Russian banks: the FinTech Association conducted 45 in-depth interviews and studied more than 100 cases of AI application in fintech