Artificial Intelligence (AI) has already permeated several areas of our lives, showcasing its tremendous potential to make a difference. While this is a positive outcome, several glaring concerns over its safety are looming. India is leading through specific safe and trusted AI initiatives through the IndiaAI mission to promote responsible AI practices and ensure fairness, transparency, and security. 

The session 'Ensuring Safety, Trust, and Governance in the AI Age', held as part of the Global IndiaAI summit, explored India's role and contribution in international forums and deliberated on balancing technological innovation with governance to protect citizens. 

S Krishnan, Secretary of the Ministry of Electronics and Information Technology, delivered the keynote address for the session. Jibu Elias, Country Lead for India, Responsible Computing Challenge, Mozilla moderated the discussion. Other panellists included Mahavir Singhvi, Joint Secretary, New Emerging and Strategic Technologies Division, Ministry of External Affairs, Government of India; Dr Balaraman Ravindran, Head, Centre for Responsible AI, IIT Madras; Timothy Curtis, Director, UNESCO office, New Delhi; Sharad Sharma, Co-founder, iSPIRT Foundation; Deepak Goel, Scientist and Group Coordinator for Cyber Laws Division and Data Governance, Ministry of Electronics and Information Technology; Dr Urvashi Aneja, Director, Digital Future Labs, and Arjun Goswami, Director, Public Policy, Cyril Amarchand Mangaldas. 

In his keynote address, S Krishnan spoke of how every technology is met with initial fears. There's much suspicion, with most people believing it could end the world. At the same time, there is also a crop of technology enthusiasts willing to go forth to any extent to bring in development and change. While the guardrails for using any technology are generally established, there is no guarantee that its ill effects will be eliminated.   

"The fears are multifold. One of the biggest concerns is whether the level of intelligence of machines surpasses human potential. There's also fear that if productivity goes up to a great extent through these machines, there will be massive job losses," remarked the Secretary.  

Jibu Elias brought the panel's attention to how the IndiaAI mission proudly advocates AI for all. "The IndiaAI mission has seven pillars, and each one is all about making it more inclusive. The idea is to make it available for everyone to innovate on. However, to make it more robust, the need of the hour is to make it a multi-stakeholder dialogue," he added.  

Ensuring AI ethics 

Dr Balaraman Ravindran spoke about bias persisting in AI by explaining how a loan manager in a bank is generally biased towards those who don't have money. The notion of bias is likely to kick in whenever there's a discriminatory decision.  "What is acceptable that aligns with our social and ethical standards? What is the notion of bias in various Indian languages? How do you express bias? There are a lot of challenges," he reiterated.  

Trust can only be developed when the process is transparent. In India, a significant fraction of people aren't well-educated and trained, so the level of explanation must go down to what someone in a village may understand.  

Dr Aneja believes that institutionalizing some process-related steps is important if India wants to advance transparency and accountability of AI systems. One such step is mandating or using data cards akin to food nutrition labels.  

In this case, end users have all the information to decide on the products they want to use and how. "We need an outcome-based approach. As a society, we define the outcomes we desire from AI products and tools, and the task for companies is to innovate towards that vision," she asserted.  

India paving the way  

Timothy Curtis highlights that what happens in India is important for the world at this juncture. Speaking about UNESCO's approach, he shared that the organization doesn't do direct research; instead, it harnesses research from around the world to foster agreements from countries of different sizes and contexts.  

In 2021, UNESCO developed the first universal recommendation concerning the ethics of AI. It was considered a ground-breaking instrument not only because it was legally binding but also because it represented a discussion among all 194 countries on the baseline ethical considerations for the development of AI.  

Partnership with industry  

Sharad Sharma believes that while the digital public infrastructure (DPI) model is appreciated and has also been mentioned in the UN General Assembly, it has its limitations. For instance, it isn't possible to achieve strategic autonomy in AI just by using this model.  

"We shouldn't think of regulation and innovation as binary things. Regulation can drive good innovation– in fact, the role of regulation is to create markets for beneficial innovation. We have enough examples of that already. That's what we should be focusing on," stated Dr Aneja.