Introduction

Principles for Responsible AI is an approach document for India towards creating responsible AI for all and has been published by NITI Aayog in February 2021.

Realizing the immense potential that AI holds, the government’s proposed national strategy on use of AI in governance areas such as healthcare, agriculture and education has brought AI in the centre-stage of the reform agenda. At the same point of time, it also underlines the need for a robust ecosystem that can solve the twin purpose of R&D as well as scaling up such solutions globally. However, despite the tremendous gain in efficiency and productivity due to introduction of AI in various sectors, the policy makers are aware of the downsides and, hence, advocating managing AI systems responsibly – a system devoid of any biasedness and prejudices. 

In the process, the approach paper explores the various ethical considerations after taking into account expert consultations and interviews with top agencies deploying AI solutions in India. Such considerations have been sub-divided into system considerations and societal considerations and have been further analyzed. While the system considerations mostly deal with the overall principles behind decision-making, rightful inclusion of beneficiaries and accountability of AI decisions, societal considerations focus on impact of automation on job creation and employment. Having said that, the paper cites that since societal impact of AI is an evolving area of research, it will be beneficial to have a long-term strategy to harvest the economic potential of it.

The paper also gives a sneak-peak into the legal and regulatory approaches for managing AI systems. Accepting the fact that the present AI ecosystem has multiple stakeholders such as private, public, research institutes, legal bodies etc., it points out the importance of laying the grounds for a common acceptable behavior among them and instant clarification in case of any dispute. There is also discussion on principles for responsible management of AI systems – principles that have been set based on the nation’s accepted value system and matches the global standards. Some of the important principles that have been considered are equality, safety & reliability, inclusivity & non-discrimination, transparency, accountability and privacy & security. 

Relevance of the Report

The paper rightly has been published at a time when the deployment of AI in India has largely grown. Nevertheless, though the intention of such deployment has been to improve access, quality and efficiency, the process have often been thwarted with risks and challenges that often has a tinge of ‘ethics & principle’ element in it. Hence, to establish broad ethics principles for design, development and deployment of AI in India, NITI Aayog, quite judiciously, has published the first part of its strategy titled “Towards Responsible AI for All” towards overcoming such challenges. It can be considered as a roadmap for adoption of AI across sectors in a responsible manner.

Key Takeaways

• There is a need for fine balancing between large-scale adoption of AI for public good and protecting societal interest
• There are two main considerations when it comes to ethical challenges: a) system considerations and b) societal considerations
• The system considerations primarily deal with challenges such as lack of understanding of AI functioning, explaining AI decisions and assigning accountability. The societal challenges are focused on impact on jobs and unrequired psychological profiling
• There are 7 broad principles that needs to be adhered to when managing responsible AI. In addition to this, it is important to see that the manner in which such principles are implemented is enabling a positive environment for adoption of responsible AI ecosystem in the country