OpenAI Introduces IndQA to Enhance AI Competency in Indian Languages
- OpenAI rolls out IndQA to evaluate AI performance across Indian linguistic and cultural contexts
- The IndQA benchmark spans 12 Indian languages and 10 culturally relevant domains, covering topics from history and cuisine to media and spirituality.
- The initiative is designed to help develop AI systems that better understand and respond to India's diverse linguistic and cultural contexts, promoting greater inclusivity and relevance.
OpenAI has unveiled IndQA, a new benchmark designed to improve how artificial intelligence systems understand and respond to questions rooted in Indian languages and culture. The initiative reflects OpenAI’s broader goal of ensuring AI benefits people across all regions and languages.
In its announcement, OpenAI emphasized that about 80% of the global population doesn't use English as a primary language. Yet, most current tools measuring AI performance in other languages are limited. Existing multilingual benchmarks have reached a point where top AI models show similar scores, making it harder to assess true progress. Additionally, many of these tools focus only on translation or multiple-choice answers, missing out on cultural context and lived experiences.
India was selected as the starting point due to its vast linguistic diversity and growing AI user base. With 22 official languages and nearly a billion non-English speakers, India is already ChatGPT’s second-largest market.
IndQA includes 2,278 questions across 12 Indian languages such as Hindi, Bengali, Tamil, Telugu, Gujarati, and even Hinglish. It spans topics like food, history, cinema, festivals, law, and sports, with inputs from 261 experts across the country.
Also Read: OpenAI Offers Free ChatGPT Go Access for Indian Users for a Year
Each question includes a native-language prompt, an English translation, expert grading criteria, and an ideal response. A model-based system then evaluates AI responses against these standards.
OpenAI stated that IndQA is a step toward making AI models more context-aware and plans to expand similar benchmarks to other regions in the future.
Read More News :
