Govt’s AI advisory regressive, clarification vague: Experts

Following a blowback from the industry, Chandrasekhar on March 4 clarified that the advisory does not apply to startups

The move to get companies to seek government permission before launching an AI product in the country has been heavily criticised by artificial intelligence experts, terming the March 1 advisory as “regressive” and one that will throttle innovation.

Experts say the advisory and certain terms used in the subsequent clarification by Minister of State for Electronics and Information Technology Rajeev Chandrasekhar on the matter, are ambiguous and vague.

“We don't have extremely powerful AI at a product level yet. Therefore, regulating something like that doesn't make sense,” Srikanth Velamakanni, Co-founder of analytics platform Fractal told Moneycontrol. Although, he said the government’s intent is right, it will only make sense to regulate the more advanced consumer-facing AI products, which are yet to come up.

“A blanket policy like that would mean they will be overburdened with requests and that will completely throttle innovation,” Velamakanni added. It will make sense for government to regulate only extremely powerful AI such as GPT 4, he said.

Following a blowback from the industry, Chandrasekhar on March 4 clarified that the advisory does not apply to start-ups.

In a post on X, formerly Twitter, Chandrasekhar said, "Recent advisory of @GoI_MeitY needs to be understood. Advisory is aimed at the Significant platforms and permission seeking from Meity is only for large platforms and will not apply to startups."

However, industry experts are still not satisfied with the advisory.

Vishwanath P B of Cyber Lex Solutions said, "the MeitY advisory on permit for deploying AI models is regressive and is like shooting oneself on the foot.” Adding that it is almost impossible for the government to implement and for organisations to “effectively” adhere to the “license raj.”

In his opinion, India should enact or bring an ordinance on the lines of EU AI Act or Singapore AI Act, keeping long-term interest in mind instead of such ad-hoc advisories/guidelines in bits and pieces that serve no purpose.

Further in the post, Chandrasekhar said the advisory is aimed at “untested AI platforms” from deploying on Indian internet. Reacting on this, several experts questioned as to who will certify whether the models are tested or not.

The implication of this advisory is akin to India farming with medieval tools instead of fertilisers and tractors, which would impact services' export capability, Ramanuj Mukherjee, Co-founder and CEO of legal education company LawSikho, said on X.

“If new AI tools will not be rolled out in India, the Indian professional class will be left behind in competition with global professionals, that would be a disaster,” Mukherjee added.

Pedro Domingos, Professor of computer science at the University of Washington, said the Indian government takes the lead in the race to the “stupidity singularity.”

Not just experts, the move has not gone down well with companies. Aravind Srinivas, CEO of Perplexity AI, called the advisory a bad move by India.

Aaron Levie, CEO of cloud storage company Box, took to X (formerly twitter), and said “this is why America can win in AI.”

The March 1 advisory has asked for explicit government permission for AI models under testing and says it must be deployed only after appropriately “labelling” the output generated. The advisory also asks all intermediaries or platforms to ensure that their computer resources do not permit any bias or discrimination or threaten the integrity of the electoral process.

Further, the advisory says that if a platform creates any synthetic content that can be used to spread misinformation or deepfake, “it is advised that such information.. is labeled or embedded with a permanent unique metadata or identifier…” This metadata or identifier can be used to identify “creator or first originator of such misinformation or deepfake”, the advisory says.

The advisory also says users “dealing” with unlawful information can be punished, and that non-compliance may lead to penal consequences for platforms, among others.