New AI models and initiatives to boost India’s tech ecosystem

Google hosted its Google I/O Connect Bengaluru 2024 event on July 17, announcing new AI models, tools, and initiatives focused on the Indian AI community. The highlight was the introduction of the Gemma 2 AI model, designed to better adapt to Indian languages, which is now accessible to all developers. Google also partnered with the Ministry of Electronics and Information Technology (MeitY) Startup Hub to support India’s startup ecosystem.

Google also revealed that over 1.5 million developers worldwide have used its Gemini models to create AI-powered tools and apps. The company emphasised its commitment to empowering the developer community by providing access to multimodal AI models with extended context windows. Notably, India hosts one of the largest developer bases on Google AI Studio, a platform for building new AI tools using Gemini.

To aid developers in creating superior tools, Google made the Gemini 1.5 Pro model, which features a two-million token context window, available to all developers in India. The Gemma 2 small language model (SLM) was also released. Google’s tokeniser, a tool that breaks down text into smaller units for AI processing, was highlighted as essential for building multilingual solutions tailored to India’s diverse languages.

Google also open-sourced its Composition of Language Models (CALM) framework, allowing developers to combine specialised language models with Gemma to create more efficient solutions. For example, a developer building a coding assistant in English could offer assistance in Kannada by integrating a Kannada specialist model within CALM.

Additionally, Google introduced IndicGenBench, a new benchmarking tool for Indian languages, covering 29 regional languages, including those previously unbenchmarked. The company also announced several partnerships and initiatives, including offering up to $350,000 (roughly Rs. 2.9 crore) in Google Cloud credits to eligible AI startups. This support aims to provide the necessary infrastructure and computational power to develop and deploy AI technologies, along with AI-first programming, curriculum, and nationwide Gen AI Hackathon to support early-stage startups.