Google I/O Connect Bengaluru 2024: Latest AI Models, Tools, and Programs to Fuel Developer Innovation in India

At Google, we've been investing in AI for over a decade, constantly pushing the boundaries of what’s possible—boldly and responsibly. We're now fully in our Gemini era, bringing the power of multimodality to everyone, expanding the types of questions you can ask with advances in long context windows, and making AI truly helpful through our proprietary and open models, along with our research advancements.

Today, more than 1.5 million developers globally use Gemini models across our tools. The fastest way to build with Gemini is through Google AI Studio, and India has one of the largest developer bases on Google AI Studio today. We're inspired by the innovative solutions Indian developers are building leveraging our AI tools. We're in the early stages of an AI platform shift, and India, with its thriving startup and developer ecosystem, is well-positioned to lead this revolution.

At Google I/O Connect Bengaluru 2024, we demonstrated our commitment to democratizing AI for Indian developers by focusing on three key AI opportunity areas we are particularly excited about in India: multimodal, multilingual, and mobile. We unveiled a range of tools, programs, and partnerships designed to support developers as they build AI solutions for India and the world. We’re working with MeitY Startup Hub to train 10,000 startups in AI, expanding access to our AI models like Gemini and Gemma, introducing new language tools from Google DeepMind India, and enhancing the software development process with AI-powered features, with a steadfast focus on responsible AI.

Enabling India’s startups to be at the forefront of global AI innovation with MeitY Startup Hub

Our commitment to supporting the growth of the AI ecosystem has been long-standing through programs like our Google for Startups Accelerator: AI First and our Build with AI event series. We’re now working with the MeitY Startup Hub to support 10,000 Indian startups in their AI endeavors. We aim for this effort to ignite innovation across the full spectrum of India’s startup ecosystem, placing them at the helm of global AI innovation.

As part of this we are:

• Supporting eligible AI startups with up to $350,000 in Google Cloud credits to invest in the cloud infrastructure and computational power essential for AI development and deployment.
• Equipping startups with AI-first programming and curriculum through existing programs like Startup School and Appscale Academy, to help them with the skills, knowledge, and mentorship needed to thrive in the AI landscape.
• Developing AI innovation programs to help the next generation of startups and developers solve real-world challenges. This includes a nationwide Gen AI Hackathon, a 3-month immersive experience in partnership with MeitY Startup Hub and Startup India, and the Solve for India Startup Bootcamp | AI Edition, to support early-stage startups tackling challenges across healthcare, climate change, agriculture, cybersecurity, and digital public infrastructure (DPI) using AI.

Expanded access to our AI models

Gemini is designed to be multimodal, empowering you to reason across text, image, video, code, and more. This brings immense possibilities. Take i-Saksham for example, a non profit organization that found Gemini in Google AI Studio could help it extract actionable insights from hours-long coaching sessions conducted in Hindi with its women trainees in a mere 60 seconds.

1 million tokens lets you analyze vast amounts of data—up to 1 hour of video, 11 hours of audio, or extensive codebases and text. We’re expanding this further. The 2 million token context window on Gemini 1.5 Pro, previously waitlisted at I/O, is now available to all developers in India. This expansion empowers you to process and understand even more information in a single request, leading to more contextual and comprehensive results.

We're also thrilled by the incredible response to Gemma, our family of open models built from the same research and technology used to create the Gemini models. We've now released Gemma 2, the next generation of open models for responsible AI innovation to all developers. Gemma 2 features improvements in performance along with significant built-in safety advancements. It's available in both 9 billion and 27 billion parameter sizes, optimized by NVIDIA to run on next-gen GPUs and also runs efficiently on a single TPU host in Vertex AI.

Gemma's tokenizer, which breaks down text into smaller units for AI processing, is particularly powerful for building multilingual solutions that understand and respond to India's diverse languages. This has been demonstrated by Navarasa, a multilingual variant for Indian languages built on Gemma.

Building on our open-source resources for Indian language solutions

The Google Deepmind India team has been focussed on enabling open-source resources to help developers build language solutions for India.

Through Project Vaani, in collaboration with the Indian Institute of Science (IISc), we have been capturing the diversity of India's spoken languages. We're thrilled to have completed Phase 1, providing developers with over 14,000 hours of speech data across 58 languages, collected from 80,000 speakers in 80 districts. With our partners at IISc, we're now embarking on Phase 2, expanding to cover all states in India spanning 160 districts.

Building high-quality language models that accurately represent India's linguistic diversity can be a complex challenge. That's why we’re introducing IndicGenBench, a comprehensive benchmark designed specifically for Indian languages. Covering 29 languages, including many that have never been benchmarked before, IndicGenBench provides a valuable resource to assess and fine-tune language models.

We're also open-sourcing our CALM (Composition of Language Models) framework that allows developers to combine their specialized language models with Gemma models. This enables the creation of more powerful, efficient, and nuanced solutions that cater to specific use cases and linguistic variations. For instance, if a developer is building a coding assistant in English, by composing with a Kannada specialist model in CALM, they may be able to offer coding assistance in Kannada as well.

Efficient on-device AI with Matformer framework

We're committed to bringing the power of on-device AI to everyone. Android is the first mobile OS with a built-in foundation model. Gemini Nano, our most efficient AI model, is designed specifically for mobile, delivering fast, private AI experiences even on unreliable networks.

To further enhance on-device AI capabilities, we’re introducing the Matformer framework, pioneered by our Google DeepMind team in India. This will allow developers to mix and match different sized Gemini models within a single framework, optimizing for both high performance and low resource consumption. This will translate to smoother, faster, and more accurate AI experiences directly on users' phones.

AI for India’s Agricultural Sector

We believe in harnessing the power of AI for social good, and our Google DeepMind and Google Partner Innovation team in India has been at the forefront of tackling global challenges like flood forecasting and healthcare. We’re going to be soon launching the Agricultural Landscape Understanding (ALU) Research API, a limited availability tool designed to make agricultural practices more data-driven and efficient.

Farmers face myriad challenges, from accessing subsidies and capital to improving yields and market access. The ALU API looks to address these issues by leveraging AI and remote sensing to map individual farm fields across India, with the potential to provide landscape insights at the farm field level. Built on Google Cloud and our extensive research, including collaborations with the Anthro Krishi team and India's digital AgriStack, the use of ALU information is already being explored by select partners like Ninjacart, Skymet, Team-Up, IIT Bombay, and the Government of India.

Streamlining Software Development and an Early Look at Our AI Agents

We announced exciting updates to further streamline software development. This include Firebase AI Monitoring in private preview, offering a Gen AI-focused dashboard for real-time insights into LLM-powered features, new integrations for Project IDX, including an early preview of Android Studio on Project IDX to quickly build native Android apps, and Checks AI Safety to help evaluate, monitor, and oversee the compliance of AI models and agents. Learn more here.

We also unveiled a glimpse into the future of AI-powered development with innovative AI agents designed to make your workflow even more efficient and intuitive. We're open sourcing Project Oscar, a reference for an AI agent that helps with open source project maintenance. This empowers maintainers to focus on what they love most: writing code, while AI handles the time-consuming tasks that often disrupt their flow. We shared an early look at an AI Testing Agent in Firebase App Distribution–now available in private preview–to save time and effort in app testing. And we also showed AI Generated UI in Flutter, an early experiment for dynamic and personalized UI creation.

New Google Wallet APIs and Google Maps Platform pricing for India

We continue to invest across our platforms to help Indian businesses and developers thrive. We’re introducing Google Wallet APIs to simplify the integration of loyalty programs, tickets, gift cards. For developers using Google Maps Platform, we’re introducing India specific pricing that is up to 70% lower on most APIs to make it even easier to build location based solutions. Additionally, we’re collaborating with the Open Network for Digital Commerce (ONDC), offering developers building for ONDC up to 90% off on select Google Maps Platform APIs.

We remain inspired by the ingenuity and passion of our Indian developer community, and we can't wait to see how you'll push boundaries and reimagine what's possible with these new tools and resources.