Telangana employs same tech to issue new ration cards that deleted 20 lakh names

The technology's algorithm went wrong, leading to the exclusion of thousands of legitimate beneficiaries from welfare schemes, including subsidised food.

BySumit Jha

Published Jan 26, 2024 | 11:00 AMUpdatedJan 26, 2024 | 11:00 AM

Telangana employs same tech to issue new ration cards that deleted 20 lakh names

The Telangana government has invited applications for ration cards through its Mee Seva Portal. However, there is a catch.

The government is employing the same technology that has — since 2016 — arbitrarily deleted 20 lakh ration cards to issue new ones.

A report, “How an algorithm denied food to thousands of poor in India’s Telangana,” by Reporters Collective said that the Samagra Vedika algorithm, used to digitally profile the state’s residents, denied thousands of people the benefits of the state’s welfare schemes.

Also read: Prof Kodandaram, editor Amer Ali Khan appointed Telangana MLCs

The process

The state has asked interested citizens to apply for ration cards through the Mee Seva Portal by the end of February.

Besides those who applied for ration cards through Praja Palana, individuals who were may also apply for the cards now.

The government received 20 lakh applications for ration cards during the Praja Palana programme.

Samagra Vedika is an integrated platform developed by the Telangana government to detect fraudulent applicants in government schemes and ensure accurate identification of beneficiaries.

The report said the algorithm went wrong, leading to the exclusion of thousands of legitimate beneficiaries from welfare schemes, including subsidised food. This happened because of faulty data and inaccurate predictions about their incomes.

Also Read: Telangana government kickstarts Praja Palana

The case of Bismillah Bee

The aforementioned algorithm resulted in individuals like Bismillah Bee, whose husband succumbed to cancer, being deprived of their rightful benefits.

Such people are said to have faced significant challenges in proving their eligibility.

The algorithm identified Bismillah’s husband Syed Ali as a car owner. An Al Jazeera investigation found that a person named Syed Hyder Ali owned a car and the algorithm got confused over the names, leaving Bismillah Bee out of the scope of the benefits.

The investigation pointed out that the use of AI in welfare schemes, while aimed at removing ineligible beneficiaries, resulted in the wrongful removal of legitimate names.

It highlighted the need for greater accountability and transparency in the deployment of such algorithms.

Even as an investigation was underway into the en masse deletion of names, the state government is going ahead with the same technology.

IT and Industries Principal Secretary Jayesh Ranjan, however, defended the algorithm.

“The AI algorithm functioned well, demonstrating efficiency in filtering spurious applications by up to 95 percent. This improvement, from the previous efficiency rate of 84 percent, is a positive development,” he told South First.

“Enhancing the accuracy in identifying and eliminating fraudulent applications ensures that legitimate ones receive prompt attention. In instances where authentic beneficiaries are inadvertently overlooked, we conduct personalised verifications. Al Jazeera highlighted a single case where a genuine applicant was excluded,” he noted.

Also Read: ‘Consider income declaration, not certificate for new ration cards’

Samagra Vedika

Samagra Vedika works by gathering data from various databases, such as electricity and water connections, house and land ownership, and old-age pension records, to determine citizens’ eligibility for different welfare programmes.

The platform uses machine learning and entity-resolution technology to uniquely identify a person’s identity.

Key features and working of Samagra Vedika include:

  • Data integration: Samagra Vedika integrates data from various government departments to crosscheck citizen details and ensure accurate identification of beneficiaries.
  • Machine learning and entity resolution technology: The platform uses machine learning and entity resolution technology to uniquely identify a person’s identity, ensuring that the right individuals receive their rightful benefits.
  • Detecting fraudulent applicants: It helps in detecting fraudulent applicants in government schemes by analysing data and flagging potential issues, such as low-income certificates being submitted by financially sound individuals.
  • Transparency and accountability: The platform aims to bring transparency and accountability to the identification of beneficiaries, as it replaces the earlier system of in-person verification, which was considered opaque and prone to corruption.

Also Read: Telangana CM plans Indiramma Committees in each village

The concerns

SQ Masood of ASEEM, an NGO working for marginalised communities in Hyderabad, told South First that contrary to Ranjan’s statement that only one person was excluded, thousands in the state were deprived of benefits.

“We have helped four people in the state to file writ petitions in the high court, and now the Supreme Court is hearing our petition regarding this matter,” Masood said.

He further said that since the inception of the technology, lakhs of ration cards were deleted.

“This technology was developed by Posidex Technologies Private Limited. I went to it eight months ago to understand how it works, and its accuracy level,” said Masood.

“The civil supplies department said the work was being done by the Information and Technology Department when I went to the vendor to know the accuracy level. However, I did not get a satisfactory answer,” he said.

Masood pointed out that the state chief secretary submitted before the Supreme Court that 20 lakh ration cards were deleted and later, 10 lakh people applied for it again and the state was considering them.

“What does it mean that the state is considering 10 lakh again? It means that your technology failed. Apart from Bismillah Bee, three other people have filed writ petitions in the high court,” Masood said.

He recommended manual verification to overcome the problem. “If you are rejecting an application, inform the applicant. Due process is being bypassed and technology is verifying the details. People are not aware for months, and they have to wait for months to apply again if their application has been rejected,” he added.

Masood said he even sent an application to the minister for civil supplies. “Like Dharani, you have constituted a committee for ration cards,” he opined.

In his petition to Civil Supplies Minister Uttam Kumar Reddy, Masood requested the digitisation of data by using the appropriate application software.

“Digitisation of beneficiary databases should be undertaken, verified by comparison with other databases like Aadhaar (UIDAI), NPR, SECC, Census, and electoral data. Seeding of Aadhaar numbers in the beneficiary data would help in eliminating duplication,” he said.

“This will result in the creation of a centralised database at the state level with clear processes for updation. The process will facilitate online addition, deletion, and updation of ration cards and issuance of new cards. States and UTs should adopt a workflow-based approach for issuance, modification, or cancellation of ration cards based on the digitised data, using either the software developed by the NIC or by any other agency,” his petition stated.

Also Read: Debts mount as welfare schemes elude Telangana’s tenant farmers

Privacy issues

However, there are concerns about data privacy and the potential for social disempowerment due to the use of AI in welfare schemes.

Critics argued that technology should not be entrusted with the task of framing and implementing policies, and human intervention should be mandatory.

Despite these concerns, Samagra Vedika has been implemented in various government schemes and has been praised for its ability to save the state government a significant amount of money by reducing fraudulent claims.

They also raised serious concerns over Samagra Vedika, calling it illegal and indecent.

The socially problematic outcomes of the Samagra Vedika project have put a question mark on large-scale AI projects. The Mozilla Foundation, a non-profit under Mozilla, questioned the data usage by Samagra Vedika and flagged privacy issues.

The investigation conducted by Al Jazeera in partnership with the Pulitzer Centre’s Artificial Intelligence (AI) Accountability Network, revealed the negative impact of such welfare algorithms, which have led to unjust exclusions and placed the burden of providing proof on the affected individuals.