"Artificial intelligence could not exist without data, and the problem is not to collect enough data but to learn how to process these data volumes," said Vladislav Onishchenko, Head of Analytical Center speaking at St. Petersburg International Economic Forum.
"Artificial intelligence needs large samples and data sets, which should be combined," he stressed. "We need to learn how to generate them given that they are owned by different companies and individuals."
"As for the requirements for data, first, they must be of high quality," said Mr. Onishchenko. "Due to various data processing technologies, which are collectively referred to as artificial intelligence, the data should be reliable by 60-80%."
Other important criteria are accessibility and compatibility, i.e. the data should be created according to the rules that are clear, similar and obvious for all participants. Thus, the uniform standards should be developed.
"In terms of state-owned data, which we largely deal with, this very idea of describing data standards, exchange rules, requirements for the entire data life cycle is the basis of the National Data Management System," the expert noted.
"The government is unlikely to pioneer technologies, this should be handled by businesses, but establishing regulatory environment, disclosing and describing the state-owned data may provide a significant impetus to the development of artificial intelligence in the country, because, according to various estimates, about 50% of the data being collected is hidden within the governmental agencies," said Mr. Onishchenko.
The expert concluded that the development of artificial intelligence would immediately face restrictions when solving complex problems in the absence of joint efforts of business and the state and without global data exchange rules.