Artificial intelligence (AI) holds great economic, social, medical, security and environmental promise. That said, ensuring responsible usage of AI is very tricky. This is because many aspects of AI disrupt the current social norms and social constructs. Industry experts say that AI systems need regular audits to ensure ethical and safe usage. Thus to make sure the safe usage of AI, especially in the government systems, Tamil Nadu is bringing out an AI policy. The policy which is in its final stage of approval will be out soon. According to the policymakers, The Tamil Nadu State’s Al Policy has been carefully crafted, after considering the pros and cons of the technology.

AI is an emerging technology which has a huge scope. However, unlike other technologies, AI is a very different technology and is highly powerful. Thus it can cause serious problems if it is used without defining the boundaries. 

 “Normally, regulations catch up with innovation. However, AI is unique. It changes so fast that you cannot catch up with the innovation. So we need a strong policy to safeguard the usage,” says Santosh K Misra, CEO, Tamil Nadu e-Governance Agency. The policy which is going to be the first safe and ethical AI policy by a government intends to set certain parameters that should be fulfilled before rolling out AI-based solutions, especially for the government. 

The need of the hour 

AI has problems of bias and ethics. Thus to make any AI-based systems successful, the solutions should be inclusive, ethical and free of bias. “Tamil Nadu is always in the forefront of technology. We would like to make maximum use of emerging technologies like AI, Blockchain, IoT etc. for the good of people,” says the state’s IT minister R B Udhayakumar about the need for bringing out a special policy for AI.

 AI systems are self-learning systems. They adapt and change. Thus a system that is free of bias initially might acquire bias on the go. For instance, an AI system for credit sanctioning could develop a strong correlation with a certain pin code number or colour of a person. This is because it starts looking at patterns as it learns and could start denying credits to a particular section of people by their geography, pin code, colour or anything else. Thus an AI system has to be regulated not only before the launch but also post-launch to ensure that the system remains fair, ethical and inclusive. “The state intends to harness the power of technologies like AI. We are working on policies for these technologies to develop a robust framework for their use. While using AI ethical considerations must be taken into account, and our policy would address this issue.” says Hans Raj Verma, Additional Chief Secretary to Government, Department of IT, Tamil Nadu. 

Major features of the policy 

The major goal of the policy is to allow harnessing the power of Al for the public good while keeping it safe and ethically compatible with human values. “Thus the policy framework recommends the use of seven parameters called DEEP-MAX which provides a path to guide the roll-out of AI solutions,” says Misra. 

The Diversity Score ensures how well the Al system is trained for diversity in race, gender, religion, language, colour, features, food habits, accent and so on Equity score checks how the system promotes equity and treats everyone fairly. The Ethics Score will tell how well the Al system preservers human ethics, values of dignity, fairness, respect, compassion and kindness for a fellow human being while Privacy & Data protection score will ensure the privacy and data protection of individuals. The misuse prevention score checks whether the system has been designed by incorporating features that inhibit or discourage possible misuse. The Audit and Transparency score is to check how good is the system to audit decisions and the Cross Geography & Cross society score helps to check the working of Al across geographies and societies, especially for the disadvantaged societies. 

The new policy will ensure that any public utility tAI based system should satisfy all these seven parameters before they are being launched out. The state has started framing the policy last year. “We had a consultation with the industry, academia and other experts of AI. We took a lot of inputs from them and did our research,” says Misra. After the launch, the new policy will become the basic building blocks for the AI-based solutions that are rolled out in government systems.