AI, Aadhaar, And Government: Finding The Right Mix

AI and large datasets go hand in hand because AI algorithms need a lot of data to learn, look for patterns, and make correct predictions. With the introduction of digital public infrastructure that supports services like Aadhaar and the Unified Payment Interface (UPI), India has been looking into how data can be used to improve government. These services have produced many different sets of data that show how diverse India’s population is. So, it is interesting to think about letting AI get ideas from these big data sets.

Since Aadhaar began in 2011, more than 1.3 billion people have signed up for it. Because of its size and how it is set up, the Aadhaar database is a great way to train AI, but this may go against Indians’ right to privacy.

Concerns are clear when it comes to letting AI learn from data sets like Aadhaar. Because the Aadhaar database is centralised, it becomes a possible target for bad people who want to get their hands on private information. Also, if AI programmes copy existing biases, there is a chance that they will unintentionally target marginalised groups and slow down social progress. To deal with these problems, India could think about giving AI basic rights to balance its power and protect the privacy of its citizens. No matter what, it is important to learn more about these problems and do a thorough investigation.

Ideas for uses
Shifting the focus to the possible benefits, the Aadhaar database would be a very useful tool for national and state governments to use when they want to find trends in the Indian people as a whole. Even though there isn’t always a straight line between the size of a dataset and how well it works, it is usually true that bigger datasets lead to better accuracy. Notably, the Aadhaar database has more than a billion entries, which is a lot of information that could be used to teach AI. Aadhaar is also the result of clear groups and formats for how the government signed people up. Aadhaar is standard and well organised, which makes it easier to analyse than most datasets, which are put together from different sources.

Some ideas have already been made for how AI based on the Aadhaar dataset could be used. The national and state governments could use Aadhaar to track how aid benefits are given out and send money to the places that need it the most. In Punjab, cops would like to combine the AI facial recognition software they already have with Aadhaar so they can catch more criminals. No matter how these schemes might be used, the benefits must be weighed against the intrusion they would have on people’s lives.

Right to be alone
When making Aadhaar, the Indian government had to think about how national identity registration would affect the Indian Constitution’s right to privacy. In 2017, the Indian Supreme Court ruled in Justice K.S. Puttaswamy vs. Union of India that privacy is a basic right because it is necessary for people to be free. In views that all the justices agreed on, they said that as technology makes it easier to spy on people’s lives, it is more important than ever to find a balance between the right to privacy and government interests.

This balance can’t be reached by giving AI models access to Aadhaar information before building strong privacy protections. Training machine learning models with simple demographic information like name, age, gender, and place of residence would already be unethical. These worries are made even bigger by the fact that Aadhaar numbers are linked to private personal information like bank accounts and SIM cards. Also, none of the people who are currently registered with Aadhaar gave their permission for their data to be used for AI when they signed up. Given how much Aadhaar is used in everyday life, it would be hard for people who are worried about their data privacy to get themselves removed from the dataset. The weak security of the Aadhaar database makes it harder for the Indian government to protect the privacy of its citizens who have signed up for Aadhaar.

The chance of being biassed
People often think that computers are fair, but AI systems are known to repeat existing biases in datasets. Machine learning uses data from the past to make predictions about the future. However, training models on biassed data can make it hard to reproduce those flaws. In the US, for example, programmes like predictive policing often unfairly target groups of colour. African Americans in the US are more likely to be reported to the police, so AI models have been known to predict that African American communities are higher-risk areas, regardless of the real crime rate in the area. Based on how India was split up in the past, similar biases could be predicted.

Integrating Aadhaar and AI is hard because of issues like privacy, security, and not being able to find and fix biases in machine learning programmes that have been taught on Aadhaar data. When putting private data into AI, it’s important to think about the possibility of discrimination based on caste, religion, or income. To stop AI from making injustice worse, there must be checks and balances. This problem is made worse by machine learning’s “black box” method, in which even its creators don’t know how decisions are made. In 2019, Apple Inc. couldn’t explain why its new credit card algorithm gave men better credit lines than women, and even Google’s own engineers don’t know what parts of a website its algorithm uses to show search results.

Worries about safety
Protecting the integrity of datasets, like the Aadhaar database, is important to keep hackers and leaks from hurting people’s privacy. Because the database is so centralised, it has already been hard to protect the huge amount of information related to Aadhaar. In the past few years, information about Aadhaar has been leaked to the public more than once. In 2018, The Tribune was able to pay INR 500 to a group on WhatsApp to get personal information using an Aadhaar number. Before AI was integrated, these problems already existed, and the Indian government was the one who knew the most about how Aadhaar worked.

Adding AI to this process would make keeping Aadhaar information safe even more of a worry. The Indian government often works with private companies on contracts to build AI technologies that work well together. Giving other companies access to the Aadhaar database for AI training increases the number of entry points, which raises the risk of third-party intrusion and weakens the security of the whole system. In 2020, when a small contractor with access to a government network broke into the system, this was a similar problem for the United States. Even if all of the current problems with the Aadhaar system are solved, the Indian government would still need a way to check on potential contractors to make sure that no new problems are introduced.

AI and the need to protect privacy
To find a good balance between AI’s benefits and privacy issues, India should write down basic rights for AI and give rules for how to use data responsibly. But Indian officials need to talk about ethical AI, especially when it comes to Aadhaar. The government agrees with the concerns raised by the National Strategy for AI report published by NITI Aayog in 2018 and has said that AI actors will likely have to follow the data privacy rules set out in the upcoming Digital Personal Data Protection Bill. However, the Ministry of Electronics and Information Technology has said that it has no plans to make specific regulations for AI because it thinks that too much regulation will slow down innovation. Even though it makes sense to want to help AI grow, letting the business grow without any rules could lead to serious ethical problems. Instead, India should think about how to control AI without suffocating the young technology by looking at what other countries have done.

In 2019, the G20 approved principles for responsible stewardship of trustworthy AI. These principles tell AI actors to support inclusivity, transparency, security, and accountability. The European Union (EU) and the United States (US) have also come up with their own rules for how AI should be used in a moral way. Using these broad principles as a starting point, India could make national rules that fit the country’s needs in AI, especially for Aadhaar integration. There’s no doubt that AI has a lot of promise to improve the lives of people in India. The government needs to look into possible AI rules so that AI can be used in a responsible way and new technologies can be used to build a better future.