The Telecom Regulatory Authority of India (Trai) recently published a set of recommendations on the framework needed to regulate artificial intelligence (AI) systems and technologies. Jatin Grover decodes the Trai recommendations, why it has taken a dim view of self-regulation, the challenges in regulating AI, and the government’s thinking on the matter.
The regulation imperative
With increasing adoption, the risks users are likely to face are also rising. As AI gains potency and independence, there are growing apprehensions on privacy, algorithmic bias, security vulnerabilities, and the likelihood of job displacement.
Further, the use of data by companies operating OpenAI platforms such as ChatGPT to train their models, without seeking permission from the data owner, are also key privacy and revenue challenges. The march of deep-fake technologies and their illegal use to dupe people also presses for the need to regulate AI. Striking a balance between innovation and safeguarding societal interests is crucial to harnessing AI’s benefits while mitigating its drawbacks. Robust regulations are essential to create a sustainable AI-powered future.
Trai’s recommendations
The regulator recommended that AI technologies posing high risk to human lives be regulated, through the formation of a dedicated body—the Artificial Intelligence and Data Authority of India (AIDAI).
AIDAI must frame regulations on various aspects of AI—including responsible use—set relevant principles, and regulate use cases based on risk assessment. Further, it must ensure that the principles of responsible AI are applied at each phase of the AI lifecycle: design, development, validation, deployment, monitoring and refinement. A multi-stakeholder body (MSB) should act as an advisory body to AIDAI. The MSB should have members from different ministries and departments, industry, legal and cyber experts, academia, and research institutes. The MSB may invite representatives of relevant ministries and departments of Union and state governments on a need basis.
Challenges
AI’s dynamic evolution and usage makes designing robust regulations difficult—balancing between innovation and essential constraints is a key challenge. Its reliance on extensive data stokes concern about data privacy and security.
Rectifying biases in algorithms to prevent discrimination is a big worry. Regulatory harmony across nations is also a big hurdle at present.
Trai view on self-regulation
In some countries, companies such as Google, Microsoft, Meta, IBM, and Apple—through a partnership model—have opted for self-regulation. Since this is can’t be enforced by a government and it is difficult to assess actual compliance, Trai has pushed for regulation by the government.
Citing UNESCO’s observations, Trai said the “self-regulatory model” that prevails prioritises commercial and geopolitical objectives over people and is not good enough.
What the government and experts are saying
The stance of the government is to not stifle innovation but to regulate the emerging technology models through the prism of ‘user harm’. India’s upcoming Digital India Act will give equal weightage to innovation and social issues. This means that the regulations will not only protect the safety of users, but also give the space to companies for developing new technologies without imposing unnecessary rules. IT minister Ashwini Vaishnaw told FE in June that “no Bill can ever address everything (on AI). What we should always think upon is what are the guardrails? What are the boundary conditions? What are the principles? Those principles we set in and then apply as per situations and circumstances.” “We don’t want to be prescriptive. We are trying to set principles and the boundary conditions,” Vaishnaw had said at the time.
With regard to the Trai recommendations on AI regulating body, certain experts have said that it will unnecessary increase the burden on the companies and startups deploying AI models. According to the Broadband India Forum, any kind of ex-ante regulations on AI technologies would restrict innovations. According to the think tank, AI and digital transformation will aid India in realising its ambitious $5-trillion-economy goal.
