Hyderabad: Artificial intelligence (AI) may not have taken over the world, but it is here to stay. The emerging technology has disrupted markets, economies, jobs, and lifestyles. It would not be unfair to say that AI has already changed how technology shall be perceived henceforth.
In India alone, AI is being deployed in schools as part of smart education classes, in agriculture to improve crop yield, in healthcare to provide virtual healthcare services, and in law and order to nab criminals before they commit crimes! The use cases are manifold. This begs the question; is it becoming necessary to regulate AI? Do we need regulation at all?
Context and growth
According to estimates, more than 60 startups were working on Generative AI in India in 2023. This does not include the multitude of startups and companies working to develop and deploy AI systems across other sectors. Neither does it account for companies who solely train datasets which are in turn used by AI startups to train their models.
Globally, the growth of AI has been astounding. It is estimated that the AI market size shall exceed USD 1,000 billion by 2030. Generative AI systems already enjoy membership of over 100 million users worldwide. All of this data translates into enormous financial resources and market capitalisation for such companies and countries.
Behind such mind-boggling numbers, however, remain concerns regarding its fallouts. Experts have highlighted concerns over AI systems’ safety, reliability, bias, and ‘black-box’ nature of operation. Even after training the data sets, pinpointing the output can be tough. It is especially hard to tackle the problem of widespread deep fakes from impersonating real people and defrauding citizens. Likewise, the problem of misinformation seems to have amplified with the use of AI.
NITI Aayog & MeitY committees
There is a lot to consider for a country of 1.4 billion. The Indian Government under the relevant ministry, that is, the Ministry of Electronics and Information Technology (MeitY) has taken a cautious approach towards regulating Artificial Intelligence. While several research papers and policy briefs have been released by the central think tank NITI Aayog on Artificial Intelligence, a comprehensive regulation is yet to be released.
In 2018, the National Strategy on Artificial Intelligence was released by the central think tank highlighting the importance of AI for the future. It emphasised research on AI for use in key sectors, such as agriculture, healthcare, education, and ‘smart cities’.
Soon, in February 2021, NITI Aayog released an approach paper on Principles of Responsible AI. Taking cues from the European Union's interpretation of ‘Trustworthy AI’, the Union government think tank encouraged discourse regarding the use of AI towards responsible and safe uses. Acknowledging the ethical and societal considerations, it laid down certain principles for deploying AI systems, such as:
∙ quality, safety & reliability,
∙ inclusivity & non-discrimination,
∙ transparency,
∙ accountability and
∙ privacy & security
Not long after, NITI Aayog released Part 2 of the approach paper, Operationalising Principles for Responsible AI. The focus here was to discuss measures in which the principles above-mentioned could be put into effect by the ecosystem while developing AI technologies, to realise the vision of ‘Responsible AI’. It undertook a cross-sectoral approach by identifying certain actions to be taken by government-private actors and other stakeholders. Furthermore, it gave insights into how the Government of India was approaching AI regulation.
While the NITI Aayog has provided policy briefs and encouraged discourse around Artificial Intelligence, standing committees have been formulated to develop guidelines on various aspects relating to AI. In 2022, the Government through MeitY established four committees to prepare reports on AI and platforms; its impact on national missions and key sectors; technological capabilities; and cyber security, legal and ethical concerns.
AI Advisories
The Indian government undertook its first concrete step towards regulating AI through an advisory published on March 1, 2024. It was issued to intermediaries or platforms to ensure compliance with their due diligence requirements under the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) 2023 (IT Rules). The purpose of the advisory was to caution platforms from hosting deepfake content which was increasingly causing harm. The advisory went a step ahead and mandated that any company working on AI/Gen AI/LLMs was required to take the government’s “explicit permission” before deploying AI systems for public use. Advisories by their nature, however, are not mandatory.
Startups, companies and investors expressed concerns about this ‘over-regulatory’ approach. The widespread anxiety was heard, and MeitY responded by withdrawing the April 1st advisory, with another advisory published on March 15, 2024.
The new advisory withdrew the requirement of seeking explicit permission, yet, emphasised that platforms and intermediaries were to ensure that no unlawful uses were performed using their platforms through AI/Gen AI/LLMs. It highlighted the importance of preserving the electoral integrity by prohibiting any use of AI which could threaten the electoral process. Users were to be informed by platforms about the terms of service, to ensure that no illegal use cases of AI surfaced.
What Next?
As India awaits a comprehensive regulation on AI, the European Union has enacted and passed into law the EU AI Act. Apart from prescribing a risk-based approach towards Artificial Intelligence systems, the EU Act provides certainty to the entire AI ecosystem. The Indian Government’s approach towards AI has been pro-innovation while balancing user safety. A cautious approach is being undertaken by the Indian government towards this transformative technology. It is expected that the soon-to-be-released Digital India Act shall encompass rules on the use of AI systems. Every emerging technology requires some form of regulation, AI it seems, is no exception.
