Lack of an official AI policy could harm the marginalised, say human rights groups

The article highlights a critical paradox in India’s rapid AI adoption: while the government promotes these technologies for efficiency, the lack of ethical oversight risks entrenching systemic discrimination against marginalized communities. Facial recognition and predictive policing tools are disproportionately targeting Muslims and lower-caste groups, reinforcing historical biases rather than eliminating them. This suggests that without regulatory safeguards, algorithmic systems will likely perpetuate and even amplify existing social inequalities rather than fostering equitable access to justice and public services. Data scarcity and biased training sets further exacerbate these disparities, effectively creating a "digital caste panopticon." Marginalized populations are underrepresented in the datasets used to train AI models, leading to erroneous conclusions in healthcare, lending, and law enforcement. Consequently, systems often prioritize the needs of privileged groups while ignoring or actively harming those on the fringes. This structural exclusion means that those least represented in data are subjected to the most severe consequences of automated decision-making, deepening the divide between the rich and the poor. This case is vital for open data advocates as it illustrates the dangers of deploying algorithmic systems without transparent, inclusive, and auditable data practices. It underscores that data is not neutral and that open data initiatives must explicitly address representation gaps and potential biases to prevent the automation of discrimination. The situation warns that merely making data available is insufficient; without rigorous ethical frameworks and community oversight, open data can become a mechanism for surveillance and exclusion rather than empowerment and equity.

Source: thehindubusinessline.com
Published on 2023-09-12