[ad_1]
From identifying non-existent schools claiming scholarship benefits to detecting similar images used by multiple beneficiaries of government schemes, the Comptroller and Auditor General of India (CAG) is using Artificial Intelligence (AI) and Machine Learning (ML) in a big way.
The CAG has showcased at least half-a-dozen such case studies showing use of these tools in a ‘Compendium on Responsible Artificial Intelligence’, unveiled at the SAI20 Engagement Group Summit held under India’s G20 Presidency at Panaji recently.
One of these case studies is about the use of AI in detecting duplicate, fake and ineligible beneficiaries of Digital Saksharta Abhiyan (DISHA), the government’s digital literacy programme. As per the scheme’s guidelines, the beneficiaries trained under the programme were required to upload their photographs.
To see whether same images or different images of the same beneficiaries, or non-human images, were used for claiming the training cost, the centre for data management & analytics at CAG, which does research and analysis for phase-1 data-driven audits, developed an intelligent model using open-source platform — Python — to automatically analyse these images. With the help of this model, the large volume of images was analysed automatically, which was not possible manually.
The compendium mentions that two images of a person were used 316 times, while two images of another person were used 187 times. Similarly, eight photographs of a person were used 78 times and non-human images were found in several cases, it shows.
“The cases detected helped in identifying the risky transactions, duplicate beneficiaries, fake and ineligible beneficiaries,” it stated.
However, these findings are subject to field validation, and the actuals will be reported after field audit, a source said, adding that the audit report will be out soon. “These are techniques being followed in phase-1 data-driven audits, using which phase-II field validations will be done by using risk-based sampling, after which audit reports are finalised,” the source said.
The CAG also used an AI & ML model to detect non-existent schools claiming scholarship benefits. “Machine learning model was developed in Python to identify suspected fake schools which claims scholarship in 2017-18, based on pre-defined risk parameters identified from the data pertaining to 2019-20. A total of 17 parameters at school/institute level were identified for the model. A set of 10 different Machine Learning Algorithms was used with two different techniques…,” it stated. “The model achieved above 92% accuracy. This helped in identifying the risky samples for the field level verification.”
“The model achieved above 92% accuracy. This helped in identifying the risky samples for the field-level verification,” it stated.
The CAG also used AI to detect ineligible beneficiaries claiming benefits under the scholarship scheme, which was aimed at welfare of marginalised communities. “The model helped in detecting fraudulent cases,” it noted.
The AI was used for identifying circular trading transactions in taxation, in which fake invoices are used for claiming input tax credit.
“Using Artificial Intelligence Algorithms, specific types of circular transactions up to 8 iterations were identified. The model was trained and tested on the selected E-Way Bill data set related to Taxation in India and several circular trading transactions/ patterns were discovered,” it noted.
An official said, “AI is being used by CAG for risk-based audit sampling for field validation… This is being used for auditing of Direct Benefit Transfer schemes.”
Sources said the CAG has used AI and ML for sampling to conduct field validation as part of its performance audit of pre-matric and post-matric schemes. This has been done using data available on the National Scholarship portal for 2017-18 to 2022-23, sources said. They said that the first phase of the analysis has already been shared with the government and the report will be finalised soon.
According to sources, CAG is also using AI and ML in the process of conducting performance audits of schemes such as PM-Kisan and Pradhan Mantri Awas Yojana (Gramin).
[ad_2]
