India should prioritise strong AI governance to ensure data sovereignty, promote indigenous innovation, and address ethical concerns.
ChatGPT may represent a point of inflection in public consciousness about the power of Artificial Intelligence (AI) and its key role in society. It is time to take stock of what society should be doing about such an all-powerful phenomenon, which may determine the future course of humanity. Such an examination is needed along many axes, but perhaps one of the most important is AI’s role in governance. The direct role of AI in our governance systems needs to be understood better. There is a lot of talk about inbuilt biases in AI and what that means for governance AI. Inadvertent racial biases in algorithms and data have been flagged in areas such as deciding parole and welfare entitlements.
Most current AI is foreign-made. Even if fed with Indian data, its frameworks and mindset as also its key controls remain foreign. This foreign dependence on the “intelligence” that would run our social and economic systems would be much more debilitating and colonising than the industrial dependence of our earlier colonisation. The new atmanirbharta (self-reliance) and swaraj (self-governance) may, therefore, have to be AI self-reliance and AI Swaraj!
For example, the Reserve Bank of India (RBI) recently called for expressions of interest from companies to develop AI-based supervisory technology (SupTech). RBI supervises a complex sector with millions of financial transactions and thousands of finance-related entities. Taking AI’s help to track and supervise all this can be a good idea. But the problem is that RBI has shortlisted seven foreign companies, one of which would eventually develop and likely run its AI operations. Relevant documents suggest that RBI may not have properly split the function into different components with a view to keeping overall control in-house. This might leave too much control in the hands of a foreign company. The system will be fed very sensitive financial data, on an ongoing basis. Such data in foreign hands is highly problematic. Most legal and contractual terms only limit the taking away of personal data. But the anonymised data provides patterns. Such exposure of India’s sensitive financial data and intelligence is a serious issue. From providing unfair financial intelligence advantages to foreign players, such data and intelligence in inimical hands could be weaponised.
These AI systems are mostly first developed in the United States (US). Their algorithms represent foreign realities, mindsets, perspectives, ideologies and interests, even if later tweaked to Indian requirements. Mostly, AI models are also partly pre-trained on foreign data, which further entrenches these biases. When applied to the supervision of India’s financial systems, what comes out of such foreign-sourced AI remains strongly imbued with unwanted biases.
As one knows from one’s personal use of social media and e-commerce, once a strong digital dependency is built, even if one later recognises some concerns about a service, it is nearly impossible to unplug it. So, it would be with governance of AI. Finding problems may not be to any avail once these systems are entrenched.
What’s the solution?
India possesses world-class AI-related technical and management skills. It does not need to rely on foreign companies and should develop its own governance AI. In fact, there are foreign central banks that employ Indian companies for their SupTech needs.
Institutions such as RBI should develop enough in-house capacity to supervise and keep control of its AI development and operations. They need to split the AI governance tasks in a manner that a diverse set of outside consultants can be employed.
A few years ago, when the European Union central bank wanted AI-based SupTech, it developed a standing roster of consultants which had European as well as foreign companies on it. It can choose a private partner on an ongoing basis per the needs and sensitivities of tasks. It further selected a European non-profit group to provide ideas and linkages with local start-ups. The rationale behind the Indian central bank resorting to wholesale foreign corporate dependency for its AI needs remains unclear. The path should be for domestic AI capabilities. India should prioritise strong AI governance to ensure data sovereignty, promote indigenous innovation, and address ethical concerns specific to its culture and values.
This article was originally published by the Hindustan Times.