SAI Russia Presented to the Parliament the Best Practices of AI Implementation in the Anti-Corruption Field

In October 2023, the Federation Council held an off-site meeting “Using digitalization tools in preventing corruption violations and using artificial intelligence in the field of anti-corruption”.

The event gathered more than 40 participants, including senators, representatives of the Prosecutor General’s Office, the Ministry of Justice, the Ministry of Digital Development, the Ministry of Internal Affairs.

Representatives of SAI Russia were invited to the meeting and presented a report on the use of digital tools in the SAI for solving state audit tasks in the field of anti-corruption.

The use of neural networks and artificial intelligence allowed the SAI of Russia to analyze more than 100 plans of federal authorities for the presence of measures contained in national anti-corruption plans.

To teach the machine to understand the request and correctly interpret the uploaded data, the initial markup of documents was performed manually.

Project participants, when analyzing departmental plans, marked activities as belonging to a particular section of the National Plan. In the future, the neural network learned to mark events independently, based on the recording context.

Performing this amount of work manually would take from 300 to 400 man-days, and using artificial intelligence, provided that it is necessary to train and configure the neural network, it would take about 20 man-days. Thus, labor costs were reduced by 15-20 times, and the classification accuracy was more than 95%.

Today, it is safe to say that the use of artificial intelligence has allowed not only to reduce labor costs, but also to minimize the human factor. In addition, using the tool, it is now possible to conduct continuous, rather than random, verification of almost any amount of information in a minimum period of time.

After evaluating the capabilities of neural networks and testing the mechanism of their operation on the first project, as an experiment, using the potential of artificial intelligence in identifying corruption-causing factors in regulatory legal acts and their projects.

Thus, the SAI, without replacing the functions of the Prosecutor General’s Office, will be able to assess the activities of government bodies for corruption.

The automated process of identifying risk zones will significantly reduce the time required for processing and analyzing documents.