Transparency International defines corruption as the misuse of a person’s position of authority or status for personal benefit. Did you realise that public monies totaling about US$2.6 trillion are stolen or embezzled annually? According to Transparency International’s assessment on the 2021 Corruption Perceptions Index, India is the 85th least corrupt country out of 180.
Corruption typically affects public employees, politicians, and government representatives, but it can also occur in other kinds of institutions, including financial institutions, enterprises, civil society organisations, etc. The use of AI and ML tools is now one of the potential ways to boost the effectiveness of the fight against corruption. Jitendra Singh, a minister of state, discussed the use of AI to fight corruption in India earlier this year.
A recent essay by Adalan AI claims that AI machines are a promising weapon in the battle against corruption because of their capacity for autonomous learning and their ability to quickly scan, examine, and analyse incredibly large amounts of data. In general, it takes a lot of time and effort for humans to spot fraud, tax evasion, bribes, illicit bids, and procurements. However, if these tasks were handled by machines, they would be finished quickly and effectively.
highly hazardous
In some nations, it can be very risky for residents, journalists, or civil society organisations to fight corruption. Additionally, due of the risks and lack of protection faced by whistle-blowers, public officials and employees of private organisations who are willing to report instances of corruption in the workplace frequently choose to remain silent.
The Transparency International released Dozorro, a piece of machine learning software, according to the Adalan AI article. The software searches through tens of thousands of bids and purchases to identify those with a high risk of corruption. Organizations from the civil society sector that keep an eye on the procurement process are actively using the portal.
Private businesses frequently have the same issues since corporate wrongdoing can damage their public image. Here, AI can also assist them in protecting their public image.
digitization of data
It is crucial to digitise the data that AI systems are educated on because their data is what makes them effective. Public services are largely computerised and paper-based procedures are gradually being phased out in many nations. Digital platforms are used to gather and disseminate open data on government activities, public procurements, tenders, bids, declarations of public officials, etc.
For the construction of AI systems that resist corruption, various sorts of data can be used. As follows:
accessible government data
gathered information
Electronic traces
data relevant to state entities or private companies
overcoming the difficulties
The application of AI is fraught with difficulties and compromises. Bias, a culture of monitoring, vulnerability to repetitive losses, and inconsistency between AI-based approaches and the systematic approach are a few difficulties. A “black box problem” may present an additional difficulty. AI algorithms used in the fight against corruption must be accurate and flexible enough to learn based on freshly acquired information in order to be effective and avoid making falsely positive or falsely negative judgements.
The veracity and accuracy of AI tools should be guaranteed in order to overcome these difficulties. which calls for human experts to look into the AI’s conclusion. Additionally, the safeguarding of basic human rights and AI ethical values should be a priority in the battle against corruption.
Even though a “black box problem” with AI cannot be completely eliminated, it can be mitigated with some work. In particular, the algorithms on which AI tools are based should be designed in a way that describes the decision-making process in detail and makes it accessible along with the final decision. Finally, the information gathered must be accurate, sufficient, relevant, and impartial.
