Source: IE
Subject: Science and Technology – developments and their applications
Context: As the proliferation of Artificial Intelligence outpaces traditional regulation, there is an urgent call for participatory governance to ensure AI aligns with public values rather than just private interests.
About Democratising AI Governance: Giving Society a Voice
What it is?
- Democratising AI governance refers to an inclusive regulatory framework that shifts power from a state-centric or private-sector monopoly to a multi-stakeholder model. The goal is to pierce the social black box—the upstream human choices about data and automation—to ensure AI systems are transparent, culturally sensitive, and democratically accountable.
Data & Statistics: The AI Landscape:
- Rapid Adoption: India’s AI market is projected to grow at a CAGR of 25-35%, significantly outpacing the development of statutory legal frameworks.
- The Skills Gap: Over 70% of the public feels they lack the technical literacy to understand how AI decisions (like loan approvals or job screening) are made.
- Risk Bearing: While private firms hold nearly 90% of AI technical IP, the public bears 100% of the socio-economic risks related to labour displacement and bias.
- Bias Incidence: Recent studies show that non-participatory AI models can have up to 30% higher error rates when processing regional Indian dialects compared to standard English.
AI’s Growing Societal Impact:
- Labor Markets: AI is automating routine tasks, threatening entry-level roles in the IT and BPO sectors.
Example: Indian IT giants have begun freezing certain junior-level recruitments as AI tools now handle basic coding and documentation tasks.
- Healthcare Access: Predictive AI is being used for diagnostics, but can prioritize certain demographics based on biased historical data.
Example: AI health-tech pilots in rural India sometimes struggle with accuracy due to lack of diverse genomic and lifestyle data from those regions.
- Finance & Credit: Automated credit scoring can unintentionally exclude marginalized communities who lack a traditional digital footprint.
Example: Fintech startups using alternate data for loans have faced scrutiny for potentially high interest rates targeting the unbanked.
- Democratic Processes: Generative AI and deepfakes can manipulate public opinion and erode trust in information.
Example: The 2024 and subsequent local elections saw a surge in synthetic media where AI-generated voices of leaders were used for campaigning.
Need for Participatory Governance:
- Detecting Emerging Harms: Diverse communities can spot biases that developers—often from urban, elite backgrounds—might overlook.
Example: Local activists in India were the first to flag that facial recognition systems often fail to distinguish between different tribal features.
- Experiential Knowledge: Governance needs the on-ground context of users to understand how a tool functions in real-world conditions.
Example: Farmers using AI-driven crop advisory apps provide vital feedback on local soil variations that global datasets lack.
- Piercing the Social Black Box: Public oversight ensures that the choice of what to automate is made ethically, not just profitably.
Example: Public debate in India has slowed the deployment of AI in judicial sentencing, prioritizing human empathy over algorithmic speed.
- Building Public Trust: Transparency in how models are trained and audited reduces AI anxiety and fosters adoption.
Example: The Bhashini project’s open-source approach has gained more trust by involving citizens in contributing local language data.
Challenges Associated:
- Technical Asymmetry: The vast knowledge gap between developers and the general public makes meaningful participation difficult.
Example: During public consultations on the Digital Personal Data Protection Act, many citizens struggled to grasp the technicalities of automated processing.
- Fragmented Ecosystems: Regulation is often siloed within specific ministries, preventing a unified participatory standard.
Example: AI in healthcare is governed by the Health Ministry, while AI in finance falls under RBI, leading to inconsistent user-protection rules.
- Corporate Secrecy: Private firms often cite proprietary secrets to avoid the transparency required for public audits.
Example: Major social media platforms have resisted sharing their recommendation algorithms with Indian researchers, citing trade secret protections.
- Infrastructure Barriers: Meaningful participation requires digital platforms for reporting and open datasets that currently don’t exist for all.
Example: Rural Indians often lack the high-speed internet required to access and use AI transparency dashboards or reporting portals.
Way Ahead:
- Institutionalized Audits: Mandate Community-led AI Audits where civil society groups stress-test systems before and after deployment.
- Targeted Literacy Programs: Launch national campaigns to move beyond basic digital literacy to AI Literacy, enabling citizens to identify and report bias.
- Open Data Infrastructure: Create secure, accessible Data Commons that allow independent researchers to verify the datasets used by big tech.
- Intersectional Coordination: Establish a cross-sectoral AI Regulatory Body that includes representatives from labor unions, academia, and linguistic minorities.
Conclusion:
To prevent AI from becoming a tool that deepens inequality, governance must move from the boardroom to the public square. By adopting a participatory approach, India can ensure that its technological future is built on democratic oversight rather than opaque algorithms. Ultimately, the goal is to redistribute power equitably, ensuring AI serves the common good and earns the trust of the society it impacts.
