Editorial 1: The legal gaps in India’s unregulated AI surveillance

Context

The legal gaps in India’s unregulated AI surveillance.

 

Introduction

In 2019, the Indian government made headlines by announcing its intention to create the world’s largest facial recognition system for policing. Over the next five years, this ambition has materialised with Artificial Intelligence (AI)-powered surveillance systems being deployed across railway stations and the Delhi Police preparing to use AI for crime patrols. The latest plans include launching 50 AI-powered satellites, further intensifying India’s surveillance infrastructure.

• Commendable: Technological integration in law enforcement is commendable.
• Raises substantial legal and constitutional concerns: It raises substantial legal and constitutional concerns.
• Dragnet surveillance: The use of AI for surveillance has global parallels, often resulting in “dragnet surveillance,” a term that refers to indiscriminate data collection beyond just suspects or criminals.
• Section 702 of FISA: As observed with Section 702 of the Foreign Intelligence Surveillance Act (FISA) in the United States, even well-intended surveillance laws can result in overreach, infringing on citizens’ rights.
• Explores legal frameworks: This article explores the legal frameworks, gaps, and concerns surrounding AI surveillance in India.
• Intersection with Constitutional Rights: It examines how they intersect with constitutional rights, particularly the right to privacy.

 

Telangana Police Data Breach

• Revealed Deep-rooted Concerns: The Telangana Police data breach earlier this year revealed deep-rooted concerns about the data collection practices of Indian law enforcement agencies.
• Access to Social Welfare Databases: According to reports, Hyderabad police had access to databases from social welfare schemes, including “Samagra Vedika,” raising questions about the scope of data being collected and the lack of transparency regarding its use.

 

Lack of proportional safeguards

• Solutions for Public Welfare and Crime Prevention: While data-driven governance offers solutions for public welfare and crime prevention, these practices must be measured against the individual’s right to privacy, as guaranteed under Article 21 of the Constitution.
• K.S. Puttaswamy vs Union of India (2017): The Supreme Court of India, in K.S. Puttaswamy vs Union of India (2017), recognised privacy as a fundamental right, extending its scope to “informational privacy”.
• Ubiquitous Dataveillance: The judgment emphasised that the era of “ubiquitous dataveillance” brings challenges that must be addressed through robust legal frameworks.
• Surveillance Infrastructure: However, the extent of surveillance infrastructure in India currently lacks proportional safeguards, leading to legitimate concerns about the implications of AI-driven data collection.

 

The Digital Personal Data Protection Act (DPDPA)

• Purpose of DPDPA: The Digital Personal Data Protection Act (DPDPA), passed in 2023, was meant to provide a framework for managing consent and ensuring accountability for data privacy in India.
• Criticism of Broad Exemptions: However, the law has been heavily criticised for broad exemptions that grant the government unchecked power to process personal data.

 

Exemptions and Their Concerns

• Section 7(g): Section 7(g) of the DPDPA waives the need for consent when processing data for medical treatment during an epidemic.
• Section 7(i): Section 7(i) further exempts the government from consent requirements for processing data related to employment, a particularly concerning clause given that the government is India’s largest employer.
• Potential for misuse: These exemptions raise red flags about the potential for misuse, especially when applied to AI-powered surveillance technologies that operate on vast quantities of personal data.
• Section 15(c): Moreover, the DPDPA introduces obligations for citizens that could further exacerbate privacy concerns. Section 15(c) mandates that citizens not suppress any material information when submitting personal data.
• Punitive measures: This provision, while intended to ensure data accuracy, could lead to punitive measures for something as simple as an outdated address or technical error in data collection systems.

 

Unbalanced Legal Framework

• Heightened Scrutiny on individual data: In short, the DPDPA places heightened scrutiny on individual datawhile offering the government broad leeway in its use and collection.
• Skewed in favour of state surveillance: Given the profound implications of AI technologies in processing sensitive personal information, the legal framework appears unbalanced, skewed in favour of state surveillanceover individual rights.

 

The approach in the West

• EU regulations on AI: India is not alone in grappling with AI and its impact on civil liberties. The European Union (EU) has enacted regulations that could serve as a useful guide for India.
• Risk-Based approach: The EU’s Artificial Intelligence Act takes a risk-based approach to AI activities, categorising them into unacceptable, high, transparency, and minimal risk levels.
• Real-Time remote biometric identification: Unacceptable risk activities, such as real-time remote biometric identification for law enforcement, are prohibited under EU law unless exceptions apply, such as searching for victims of serious crimes or responding to imminent threats.

 

India’s Deployment of AI in Policing

• Lack of legislative debate: In stark contrast, India has begun deploying AI-powered facial recognition technology and CCTV surveillance in public spaces with little to no legislative debate or risk assessment.
• Example of Delhi and Hyderabad: For example, Delhi and Hyderabad have integrated AI into policing without any publicly available guidelines on how data is collected, processed, or stored, or how potential abuses of the technology will be prevented.

 

Regulatory Void in India

• AI Remains unregulated: As of now, AI remains largely unregulated in India.
• Promise of Digital India Act: In 2022, the government promised that AI technologies would be regulated under the upcoming Digital India Act, but draft legislation has yet to materialize.
• Risks of unregulated AI: This regulatory void leaves citizens vulnerable to the risks associated with AI-powered surveillance, including infringement of privacy, discrimination, and data breaches.

 

Comparative Legislation on AI

• Global trends: Countries such as the United States and members of the European Union have already begun to legislate on the use of AI in public systems, with clear categorisations and restrictions for technologies that could pose a significant threat to civil liberties.
• Absence of legal framework in India: The absence of a similar legal framework in India is troubling, especially given the government’s ambitious plans to expand surveillance capabilities.

 

Constitutional Concerns on AI Surveillance

• Fundamental constitutional questions: At its core, the debate over AI surveillance in India touches on fundamental constitutional questions.
• Right to Privacy: The right to privacy, as enshrined in Article 21, and the principle of proportionality, as outlined in the Puttaswamy judgment, demand that any intrusion into personal data be backed by law, pursue legitimate aims, and be proportionate to the goal pursued.
• Stretching of Constitutional Principles: However, the existing surveillance framework, bolstered by AI technologies, appears to stretch these principles to their limits.

 

Address the impact on civil liberties

• Issue with unchecked application: It is not the use of AI in governance itself that is problematic, but rather its unchecked application without sufficient safeguards.
• Need for comprehensive regulatory framework: A comprehensive regulatory framework that addresses AI’s implications for civil liberties is urgently needed.
• Public disclosure of cata Collection: It would help protect public interest in consonance with the ‘Right to Privacy’ if such a framework includes provisions for transparent data collection practices, where it is publicly disclosed, what data is being collected, for what purpose, and how long it will be stored.
• Narrow and specific exemptions: Furthermore, the framework must ensure consent gathering mechanisms have narrow and specific exemptions for processing data with independent and effective judicial oversight.
• Transparency in consent gathering: This will not only ensure transparency in consent gathering but also safeguard the constitutionality of such applications of AI-based data processing.
• Adoption of EU approach: In this context, India could benefit from adopting a risk-based regulatory approach, such as the EU’s, which categorises AI activities based on the risks they pose to citizens’ rights.

 

Balancing AI and Constitutional Rights

• Crucial juncture in AI surveillance: India is at a crucial juncture in deploying AI-powered surveillance.
• Balancing technologies with constitutional rights: While integrating advanced technologies in law enforcement and governance offers immense potential, it must be balanced against citizens’ constitutional rights.
• Privacy measures in infrastructure: Policy decisions that embed privacy measures into infrastructure before deployment, with inherent safeguards in surveillance protocols, are vital.
• Importance of pre-deployment safeguards: Consent mechanisms, transparency reports, and judicial oversight at relevant stages of data collection and management can avoid costly retrofits and retraining.

 

Conclusion

Though the DPDP Act addresses some issues, criticisms persist, and the long-awaited DPDP Rules remain unnotified. To mitigate risks from AI-driven surveillance, regulating “high-risk activities” through restrictions on digital personal data processing and transparent auditor oversight of data sharing is crucial. A proactive regulatory approach will ensure AI serves public interest without compromising civil liberties.

---

Editorial 2: Stuck in the classroom — students, teachers, NEP 2020

Context

Increased classroom time runs the risk of students becoming passive recipients, affecting the vision of the NEP 2020.

 

Introduction

Indian students in Higher Education (HE) are spending considerably more time in the classroom than their European Union (EU) and North American counterparts. Yet, they remain at risk of being relatively undereducated. There are primarily two reasons: higher proportion of teaching time in course credits and higher number of courses a semester under the National Education Policy (NEP) 2020.

 

A contrast and the academic impact

• EU and North American students: An average student in a university in the EU or North America takes approximately four courses a semester with a maximum of three hours of lectures per course a week. This brings the total classroom time to a maximum of 12 hours a week.
• Indian students: On the other hand, Indian students enrolled in the new four-year undergraduate programmesin Indian universities must take five courses a semester with four hours of lectures per course a week. This amounts to 20 hours of classroom time a week.
• Limited time for academic activities: These extra eight hours in the classroom do not leave much time for essential academic activities outside the classroom such as self-study, reading, or working on assignments, most likely leading to exhaustion and reduced learning.

 

Assessment Challenges

• Reduction in assessments: A casualty of this increased classroom time is the number of assessments that are feasible in a course.
  • In the earlier version of the choice-based credit system in the three-year undergraduate programme, where students took only four courses a semester, there was relatively more scope for continuous assessment.
  • Now, with increased classroom time, students find it difficult to work on anything more than two assessments per course.
• Impact on assessment diversity: This could impact the diversity of assessments, privileging multiple choice questions-based assessments that are easily graded via phone apps over assessments such as a term paper or a reflective essay that requires more time and effort from students.
• Incentivising rote learning: Thus, increased classroom time risks incentivising rote learning and perpetuating the school dynamics where teachers are owners of knowledge and students are passive recipients.
• Pushing students to own learning: At least at the university level, students need to be pushed to own their learning.
  • This is possible only if they are allowed time to reflect, plan, and execute their learning, explore learning outside the classroom individually and with peers.
  • This exploration can be scaffolded by assignments such as reflective essays, group projects, and cross-disciplinary problem solving.

 

The subject of continuous assessment

• NEP 2020 and continuous assessment: Addressing the reduction in the number of possible assessments is important because NEP 2020 lays emphasis on continuous assessment.
  • In this system, the final grade can be aggregated from three or four assessment components spread over the semester.
• Benefits of continuous assessment: Such a system provides an opportunity to design a mix of low and high stakes assessments, incentivising continuous effort and learning, rather than cramming up before one or two examinations.
  • Flexibility for faculty: Continuous assessment allows considerable flexibility for faculty to tailor assessment frequency and type to meet the learning outcomes of their courses.
  • Feedback loop: It is also a way to receive continuous feedback for faculty to adjust teaching strategy and for students to adjust self-study strategies.
• Extra classroom hours for Indian teachers: The extra eight-hour a week in the classroom for Indian teachers seats into the time available for research, course revisions, development of new courses, and cross-disciplinary collaborations.
  • This negatively affects the quality and currentness of teaching.

 

Comparison of Teaching Load in India vs. EU/North America

• Teaching load in EU and North American universities: The classroom time of two to three hours a course a week in the EU and North American universities, with a total teaching load of two to three courses a semester brings the average weekly classroom teaching load of a typical university teacher in these countries to nine hours.
• Teaching load in Indian Universities: In contrast to this, an average Indian faculty is expected to teach 14-16 hours a week, with time spent in the classroom varying from eight-16 hours depending on how flexible the institutional administration is in interpreting University Grants Commission guidelines.

 

The centres of learning

• Role of teachers: Teaching a course as per the vision of the NEP 2020 includes:
  • Designing the course.
  • Selection of reading materials.
  • Development and administration of assessments.
  • Grading.
• Contrast with earlier model: This is in complete contrast to the earlier model where teachers were responsible mostly for classroom lectures with assessment and grading taken care of centrally by the affiliating university.
• Elite central universities and institutes: The elite central universities, Indian Institutes of Technology (IITs), and the Indian Institutes of Management (IIMs) could be an exception to this, with possibly fewer than eight hours a week in classroom teaching per faculty along with substantially higher resources.
• Bulk of teaching and learning: It is important to note that the bulk of teaching and learning in India happens in public universities and colleges, and not in these elite institutions.

 

Conclusion

Thus, to realise the vision of the NEP 2020 fully, a serious reconsideration of the number of courses and classroom time a course in the new four-year undergraduate programmes across India is necessary. Doing so will improve the teaching and learning outcomes for Indian students putting them on a par with their global counterparts. It will also get students out of the habit of rote learning, improve their self-learning skills, and ensure their readiness for further higher-level educational pursuits.

---