The promise of generative AI for suicide prevention in India

Abstract

The World Health Organization (WHO) estimates a global suicide rate of 9 per 100,000 people, amounting to 720,000 preventable deaths each year. Despite concerted multisectoral efforts, suicide prevention remains a complex public health challenge, shaped by the interplay of socioeconomic, cultural and stress-related factors. In India, the decriminalization of suicide via the 2017 Mental Healthcare Act¹, the formation of the National Suicide Prevention Strategy in 2022, and the introduction of the Tele MANAS (Mental Health Assistance and Networking Across States) program represent important strides in the right direction. However, implementation often remains inconsistent across states². At this juncture, generative artificial intelligence (GenAI), particularly large language models (LLMs), may potentially empower multisectoral suicide prevention efforts, particularly in resource-constrained settings.

Most Indians who die by suicide do not show a history of depressive or psychiatric disorders; instead, they report intense feelings of loneliness³. Stigma and myths surrounding suicide often prevent timely help from familial or social support systems. Suicide helplines in the country, the first safety net, have been reported to be run by untrained volunteers², incapable of providing adequate support. In the digital sphere, most suicide prevention apps remain insensitive to the cultural and linguistic diversity⁴ of the Indian demographic, and offer little for marginalized groups (for example, Dalits, LGBT+ individuals) and people with lower incomes. Given that most Indians are non-English speakers with limited access to technology, digital interventions need to overcome the trinity of Indian challenges — affordability, accessibility and multilingualism. For solutions to add value, emphasis must be on developing cost-effective models that can be deployed offline or in a hybrid mode, aided by native-language⁵, audio-first support.