Education

How Artificial Intelligence Fights Corruption and Controls the Government

How Artificial Intelligence Fights Corruption and Controls the Government

Learn: The Philosophy of Artificial Intelligence

Learn more

The system was developed by Irio Musskopf, a 23-year-old expert in the field of data analysis. He brought together nine like-minded people, including programmers, scientists, designers, and journalists. To ensure the project's complete independence, the team decided not to seek funding from government agencies, but launched an online crowdfunding campaign. This decision allowed them to retain control over the project and attract support from interested users.

The project received widespread support from Brazilians, who contributed their donations and helped raise the required $25,000 in a short period of time. The "Serenade of Love" team expanded significantly, reaching over 700 volunteers. Initially, the creators planned to work on the project only in the evenings and on weekends, but they soon became so engrossed in the work that the project became their main occupation for many years to come.

A working system was developed in a record 90 days. The "Serenade of Love" project was officially launched and presented to the general public on September 7, 2016, coinciding with the celebration of Independence Day in Brazil. This launch was a significant event, highlighting the rapid development and innovative approach to the project.

The developers explained the choice of the name of their system by stating that they sincerely love their country and consider this program the best way to express these feelings. The name Serenata de Amor is also associated with the meaningless names of Brazilian police operations against criminal groups, reports of which often appear in local news.

A key element of the Serenata system is the robot program "Rosie", designed to automatically analyze the financial reports of Brazilian parliamentarians. This tool allows for the effective monitoring and evaluation of the financial activities of public servants, promoting greater transparency and accountability in budget management. Using Rosie helps identify potential violations and anomalies in reporting, making it an important tool for the fight against corruption and improving government control.

Rosie's artificial intelligence effectively supports the project's volunteers in monitoring suspicious expenses. On average, out of one million documents analyzed, the system identifies three thousand with possible signs of legal violations. This makes the volunteers' work more focused and effective, ensuring the transparency of financial transactions and helping to combat fraud.

When Rosie detects suspicious data, she posts a standard message on the official @RosieDaSerenata account: "Questionable expenses detected by a member of parliament. Help me verify this information." The message includes a link through which any Brazilian citizen can view the financial report that attracted the AI's attention. This helps increase the transparency of financial transactions and engage the public in the process of monitoring member of parliament's expenses.

The program does not have the ability to independently file complaints with government agencies. This function is performed by a person – a Brazilian citizen. Typically, the initiative is taken by volunteers who organize their activities in a Telegram channel. Such communities actively work to protect citizens' rights and ensure transparency in public administration. Volunteers play a key role in collecting information and coordinating actions for filing complaints, which helps improve public oversight and increase government accountability.

Rosi uses machine learning algorithms to identify suspicious financial transactions. This process is known in the scientific community as data anomaly detection. For example, a significantly high compensation amount might be considered an anomaly. In this case, the member of parliament requested a refund of 122 reals spent on food, which raises questions, as his previous food expenses had never exceeded 77 reals in a single day. These technologies effectively monitor and prevent financial irregularities, ensuring transparency and security of spending.

The Rosi system tracks expenses made outside the country or in Brazilian cities where the member of parliament is not required to be for official reasons. Purchases made by members of parliament from companies owned by their close relatives are also monitored. This allows for the identification of potential violations and abuses of office, ensuring transparency and accountability in the actions of public servants.

One report revealed that a member of parliament requested budgetary reimbursement for 13 meals he claimed to have eaten in one day. Another case documented an attempt to reimburse the cost of beer purchased during a vacation at a Las Vegas casino. It's worth noting that in Brazil, officials are prohibited from requesting reimbursement for alcoholic beverages. These incidents highlight the importance of controlling the spending of public servants and the need to comply with legislation on budget expenditures.

Frame: Avito / YouTube / Skillbox Media

During its operation, the system has identified 8,000 violations among 700 officials, including two presidential candidates. R$3.6 million has already been returned to the budget, and this figure continues to grow. The benefits of the Serenata de Amor project significantly exceed the funds donated by Brazilian citizens for its development. This initiative demonstrates the effectiveness of using technology to combat corruption and increase transparency in public administration.

The project adheres to principles of complete openness. The Serenada program code and documentation are hosted on the GitHub platform. The developers are confident that their work will be useful not only in our country but also for activists from other countries. In this regard, the free Serenata Toolbox, designed for use with the popular Python programming language, was introduced. This set of tools helps simplify the project's work and expands the capabilities of developers, allowing them to effectively apply solutions to a variety of tasks.

The project team is actively working to improve the Serenata system, expanding its functionality to analyze financial reports not only for deputies but also for all officials, including governors and municipal managers. This update also plans to improve the Rosie algorithms, which will identify new types of anomalies. These changes will help the system more effectively detect even the most complex attempts at corruption and misappropriation of budget funds.

Public Oversight Everywhere

"Serenada of Love" is not the only project aimed at combating corruption in Brazil. Dozens of similar initiatives are currently being launched across the country to promote transparency and accountability in public institutions. These projects unite the efforts of citizens, organizations, and government officials to create a more honest and just society.

The "Politician Profile" website (Perfil Politico) provides comprehensive information on candidates seeking various public offices, including parliamentary and presidential positions. Users can find information about previous elections in which candidates participated, as well as review their performance in previous positions. This access to information helps voters make informed decisions and better understand the political situation in the country.

The Parlametria project uses machine learning and data analytics to collect and visualize information about the activities of both chambers of the Brazilian Parliament. This tool allows access to important data, analyze legislative initiatives, and track vote results, thereby increasing transparency and understanding of public institutions.

Data on parliamentary work is often presented as numerous, anonymous documents on official websites, making it difficult to understand. The Parlametria AI project addresses this problem by extracting valuable information from documents and organizing it in a convenient and understandable format. This allows users to easily find the information they need about parliamentary actions and decisions, promoting transparency and accessibility of legislative data.

The system's algorithms analyze the most debated proposals in Parliament. The program categorizes bills into categories, such as health care and education, and also evaluates the level of debate on each one and the degree of member engagement. All collected data is presented on the website as clear infographics, allowing users to easily digest the information.

Parlametria's artificial intelligence analyzes each bill and each member of parliament, calculating a value known as "temperature." This temperature reflects the level of attention and activity of members of parliament toward a specific proposal. The higher the temperature, the more actively the bill is debated, which may indicate its importance and likelihood of passage.

The "Caring for My Neighborhood" (Cuidando do Meu Bairro) program monitors all projects financed by the São Paulo city budget. This initiative aims to increase transparency in the use of budget funds and engage citizens in monitoring the implementation of local projects. Thanks to the program, residents have the opportunity to track which initiatives are being implemented in their neighborhoods, as well as participate in the discussion and evaluation of their effectiveness. "Caring for My Neighborhood" improves the quality of life in the urban environment and strengthens the relationship between government and the population.

When road repairs, streetlight replacements, or tree plantings are carried out in the city, information about these works is displayed on an electronic map available on the project website. This map allows residents and visitors to the city to track current changes and planned works, ensuring transparency and awareness of the improvement of the urban infrastructure.

Every resident has the opportunity to learn about current and planned works in their neighborhood or in the immediate vicinity.

Local residents have the opportunity to report improperly performed works by sending complaints to the authorities through a special form on this website. This method is convenient and, importantly, effective for solving problems in the field of public services and amenities.

Social connections predict corruption

Brazilian data science specialists have presented an interesting approach to monitoring and predicting corruption crimes. In their study, they analyzed all reports of political corruption scandals published in the local press over a 27-year period from 1987 to 2014. This analysis helps identify patterns and trends in corruption offenses, which can significantly help develop effective anti-corruption strategies. Combining media data with modern analytical methods opens new horizons for increasing transparency and accountability in the political sphere. Based on these publications, the researchers created a social graph in which nodes represent businessmen, officials, and politicians implicated in corruption schemes. Connecting lines illustrate the social ties between these figures, making it possible to identify their interactions and influence within the framework of corrupt practices. As a result, a network consisting of 404 nodes and 3,549 connecting edges was created. This graph clearly illustrates the relationships between people and also shows the level of their involvement in 65 corruption cases studied. Data analysis reveals important aspects of corruption schemes and allows for a better understanding of the dynamics of interactions between participants.

Each node represents a person, and edges between them are formed when two people are involved (at least once) in the same corruption scandal. Image: Haroldo V. Ribeiro / Journal of Complex Networks / Oxford University Press, 2018

The algorithms successfully identified true super-corrupt individuals—key figures involved in multiple criminal schemes simultaneously. Nodes associated with these individuals were significantly larger than nodes associated with other participants in the corruption cases, demonstrating their high degree of involvement and influence in corruption networks.

Analyzing corruption networks reveals numerous details about illegal schemes whose organizers often go undetected. Graph studies have shown that most corruption cases involve small groups of no more than eight participants. Furthermore, a correlation has been observed between increased publications on the topic and four-year election cycles. These factors highlight the importance of further studying corruption networks to identify and combat illegal activities.

By examining social connections, data scientists have been able to predict which individuals are most likely to engage in future crimes. Various algorithms were used to predict missing connections based on the similarity between nodes in the network. These methods make it possible to identify potential threats and develop strategies to reduce crime, making data analysis an important tool in the security field.

The researchers confirm that their method for creating and analyzing corruption graphs can be effectively used to identify criminal patterns in various countries facing similar problems. This method provides tools for in-depth analysis and visualization of corruption connections, which contributes to a more effective fight against corruption globally.

On our blog, you will find a variety of interesting and useful materials to help you expand your knowledge and skills. We regularly update our content to offer relevant information and useful tips on a variety of topics. Don't miss our latest articles to help you understand the most pressing issues. Stay tuned and follow our new publications to stay informed about the latest trends and innovations in your field of interest.

  • "Data Science is a synthesis of all my interests"
  • OneHalf: The History of the Computer "Coronavirus"
  • How to Start Programming in Python

Philosophy of Artificial Intelligence

You'll get answers to important questions about "machine thinking" and understand the role of philosophy in the development of AI. You'll identify the key differences between natural and machine intelligence, and be able to conduct relevant scientific research and present in front of an audience.

Find out more