As the Summit of the Future unfolds at the United Nations General Assembly in New York, the governance of Artificial Intelligence is making headlines. The UNGA has adopted the Global Digital Compact, a blueprint of shared values to determine how technologies like AI should be governed globally. However, like the conversation on AI ethics, translating these high-level values into practical implementation frameworks is left to lawmakers across the world.
The Compact calls for a “balanced, inclusive and risk-based approach” to the governance of AI, as well as the “full and equal representation of all countries” in this process. Such lofty ideals, reminiscent of Sustainable Development Goals, are hard to realise. India may, therefore, do well to weigh its interests in AI carefully, before committing to a path.
India has the following options at the moment: enact a wide-ranging law like the Europeans, follow a more sector-specific approach like the Chinese or Americans, or advance soft law in the governance of AI. With the caveat that these debates are mostly linked to the integration of AI in digital services, and not robots or autonomous vehicles, let us evaluate each in turn.
The European folly
The most visible and debated example of a horizontal law is the European Union’s AI Act, which sets rules for different AI systems. The law is “risk-based” – a value the Global Digital Compact is in favour of. For instance, AI with unacceptable risks, like social scoring, is banned and high-risk AI systems require prior approval.
But the EU has a track record of regulating first and asking questions later. Since 2016, the EU member states have enacted around 100 tech-focused laws and tasked over 270 regulators. Europe’s own tech companies are, in fact, among the most vocal critics of the compliance burdens linked to these regulations. AI’s potential as a transformative technology must lead us to favour a more calibrated approach; one that doesn’t stem from a compliance mindset or obsession with regulatory jargon.
Also read: Digital Competition Bill can stifle opportunities for MSMEs while aiming for a fairer market
Narrower alternatives
Various laws in China now focus on specific activities like algorithmic recommendations, and synthetic content generation or deepfakes. Predictably, these add another layer of control in an already suffocated information ecosystem. Meanwhile, California’s contentious new AI safety bill has also ignited fierce debate on whether narrow laws are the best way to regulate AI. The bill imposes strict obligations on developers to implement safety measures and conduct audits on frontier models. Greater state control is not necessarily a virtue worth emulating.
Critiques of the California Bill echo concerns with Beijing’s approach—such laws can be overly restrictive, without requisite stress testing. India must tread carefully here as technology or market-specific laws once made take a long time to be replaced. The Telecom Act passed last year followed the Telegraph Act of 1885. Surely, we cannot afford to wait for decades to correct if we get our prescriptions wrong?
Also read: MeitY advisory on social media safety can backfire. It must be revisited
Soft laws and standards
India should leverage technical standards, as they are usually crafted by experts, and leave scope for agility. Standards not codified into law can adapt more quickly to a rapidly evolving AI landscape. The International Organization for Standardization (ISO) is already leading the way in standards linked to risk management and impact assessments of AI. The Bureau of Indian Standards (BIS) is moving along similar lines.
But engaging the private sector, and civil society in standard-setting is critical. The Telecom Engineering Centre attempted standards for AI fairness without sufficient focus on such an approach in 2022. It received criticism for a lack of pragmatism and was consequently shelved.
What can India do next?
India has the opportunity to push for practical frameworks at the global stage, and technical standards exemplify these. The country has already made a push for international standard-setting in telecom via homegrown institutions such as the Telecommunications Standards Development Society, and is beginning to take standard-setting in services more seriously.
Additionally, public sector AI deployments, particularly in essential infrastructure like airports and public services, require greater attention. Joe Biden’s 2023 Executive Order on AI in the United States is a useful point of departure. The Order directs federal law enforcement to work with the executive to ensure emerging technologies such as facial recognition do not erode civil rights. Measures like these can ensure that citizens are protected when AI is embedded in public infrastructure.
Deployment of AI by the state must be subject to stringent safeguards—mere values are of limited importance where the AI deployer creates an asymmetric balance of power. There is also an opportunity for India to show that emerging economies can be considered in the deployment of new technologies. They can build public sector competencies to use AI safely and build sandboxes to test use cases before unleashing AI systems on citizens. This will be much more of a global public good than a digital compact.
The authors work at Koan Advisory Group, a technology policy consulting firm. Views are personal.
This article is part of ThePrint-Koan Advisory series that analyses emerging policies, laws and regulations in India’s technology sector. Read all the articles here.
(Edited by Zoya Bhatti)