Oppenheimer, a cinematic marvel by Christopher Nolan was released worldwide in theatres on July 21. While theoretical physicist J Robert Oppenheimer juggles between morality (human rights) and national security, the audiences are compelled to see how technology in the name of security can unleash an insurmountable power in the hands of the state. Similarly, serious concerns are being raised by entrepreneurs, scientists, and political analysts concerning Artificial Intelligence (AI), a kind of superintelligence more powerful than any other technologies of the past. On July 18, the UN Security Council convened its first-ever debate on AI to access its potential impact on global security and the necessity to regulate it. For that, it is important to understand the primary role data plays in shaping AI. It has potential lessons for India’s prospective Digital Personal Data Protection law.
AI gives machines the ability to learn from experiences and perform cognitive tasks, which were once the forte of humans. AI promises it can do everything ranging from self-driving cars (Google Car project Waymo), making human-like conversations through tools (Chat GPT, Dall-E, Midjourney, Codex, etc.), controlling our gadgets and shaping our choices. AI is also being used by governments to control its people, facilitate social welfare, decision-making, criminal tracking, etc. AI is created by the amalgamation of data and algorithms (set of instructions) which is fed to software. The software learns from patterns and features. It requires a massive amount of data to understand and improve decision-making processes thereby turning data into intelligence.
Data is not simply facts or information stored in a computer or network; it is ontological as it can independently exert causal force or produce effects, in this case for AI. Data is not neutral. It has racial, ethnic, caste, and gender prints. At the international level, there is a unidirectional flow of data from the Global South to the Global North leading to digital colonialism (Facebook has most users in India but the data centres are located in the US and Europe). At the domestic level, data gives power to governments to carry out surveillance, infringe on the rights of their people, and undertake targeted killings based on ethnicity, religion, gender, caste, etc. — all under the garb of national security. For example, actors linked to Myanmar military and radical Buddhist nationalist groups systematically used Facebook’s algorithms based on user data to incite the targeted killing of Rohingyas; China is using data based on people’s online chat history, social media posts, medical records, bank accounts, its AI-based facial recognition technology for mass surveillance in Xinjiang and Tibet; Cambridge Analytica trained on Facebook’s data was later misused by the US government, etc. Hence, riding on the back of data, AI is facilitating the violation of dignity, freedom and equality.
This is the more significant threat to humanity. Thus, before the efforts are made to create responsible AI laws and regulations, the laws around data, that is, keeping personal data safe and upholding people’s right to privacy have to be reconciled. This has been showcased by European Union as it had laid the strong foundation of data protection centred around the rights of its people in the form of General Data Protection Regulation before inching towards a responsible and safe AI Act.
In this background, it is important to delve into the upcoming data protection regulation in India. As the Monsoon Session of Parliament comes to a close, an important Bill called the Digital Personal Data Protection Bill awaits to become law (it has been passed in the Lok Sabha). The Bill aims to regulate data which is personal and private. It will set the tone for India’s AI governance as well. The Bill primarily addresses the handling of digital personal data, acknowledging the importance of safeguarding an individual’s personal information while also recognising the necessity of processing such data for legitimate reasons.
