Digital India Bill may bring in guardrails against AI algorithms, offer opt-out rights
The bill may make it mandatory for bodies to explain to citizens the rationale behind a decision taken by an algorithm, and how it arrived at the decision by processing of which particular user data.
In the upcoming Digital India Bill, the government is likely to bring in provisions bolstering algorithmic accountability, including empowering citizens to opt from being subjected to an algorithm's decisions, albeit with riders.
This comes at a time when artificial intelligence-based algorithms have often been categorised of being as discriminatory, promoting bias and so on.
The government had earlier expressed its intention of bringing in algorithmic accountability in future legislations, taking into cognisance the immense rise in deployment of artificial intelligence in everyday life, including governance.
The upcoming Digital India Bill will replace the over-two-decades-old Information Technology Act and provide comprehensive oversight mechanisms over the digital landscape.
How the opt-out mechanism may work
According to information available with Moneycontrol, this and similar provisions may be a part of Digital India Bill in a separate section on the rights available to a citizen for safeguarding themselves against a decision made by a technology.
If a citizen chooses to opt-out of being subjected to algorithmic decisions, the government may bring in caveats, including foregoing any corresponding digital service which cannot be made without that algorithm's decisions.
The opt-out mechanism may also not be available if the technology-based decision is required or authorised by law or if it is permitted under any law.
Human oversight over algorithmic decision
The bill may make it mandatory for bodies to explain to citizens the rationale behind a decision taken by an algorithm; explain how it came to the decision by processing of which particular user characteristic and the role it played in the decision making.
According to information available with Moneycontrol, in certain situations where citizens could not opt out of algorithmic decisions, they may be able to request an intermediary to do a human review and reconsider the decision.
In that particular case, the user is likely to receive the reconsidered decision that has been reviewed by a human along with explanation in a "reasonable amount of time".
These provisions are likely to have a large impact on content moderation by social media platforms, as they are heavily automated with the help of artificial intelligence.
Social media platforms such as Meta, X (formerly Twitter) and others, also have human reviewers, and the AI sends particular cases to be reviewed to such reviewers.
If the government brings in provisions that require human oversight of decisions made by algorithms, it may require a massive overhaul of a platform's capabilities and personnel.
Bill expected to be out soon
As of now the government has conducted pre-consultations on the Digital India Bill in Bengaluru and Mumbai to explain the principles of the upcoming legislation.
During those discussions, Minister of State for Electronics and Information Technology (MeitY) Rajeev Chandrasekhar said, "The new law should evolve through rules that can be updated and address the tenets of Digital India which are open internet, online safety and trust, accountability, adjudicatory mechanism, and new technologies."
During these meetings, Chandrasekhar had said that the government was considering removal of the safe harbour provision for internet intermediaries in the bill.
The safe harbour provision gives internet platforms legal immunity against content shared by users on the platforms and was allowed by the IT Act, which the forthcoming Digital India Act seeks to replace.
