Generative AI: Take the lead Bengaluru, for India

Nvidia, Adobe, Google and Microsoft have all made significant announcements on Generative AI following the release of GPT-4 in March. There is a flurry of activity in Indian industrial circles to determine our role and ability to lead in areas of Generative AI.

In my heavily biased opinion, Bengaluru can lead the charge for India in Generative AI.

With its robust talent pool and vibrant start-up ecosystem, Bengaluru has positioned itself as a formidable global AI player. According to the HBR TIDE ranking, which evaluates the top 50 AI cities, Bengaluru secures an impressive fifth position, with only San Francisco, New York, Boston, and Seattle ahead. Bengaluru boasts of the world’s second-largest AI talent pool, showcasing its commitment to fostering an environment conducive to AI-driven advancements. Additionally, it ranks fifth in diversity among AI workers, making it a thriving hub for talent from diverse backgrounds.

The start-up ecosystem in Bengaluru is another significant strength, fuelling innovation and providing a fertile ground for disruptive ideas to flourish. Start-ups in Bengaluru are at the forefront of harnessing the transformative potential of Generative AI, redefining industries, and driving substantial impact.

This vibrant start-up culture, and a supportive ecosystem, make Bengaluru an ideal hub for cutting-edge Generative AI applications.

Also Read | Artificial intelligence's potential in enhancing India's tech ecosystem vast: PM Modi

During the recent CII Knowledge Summit on Generative AI, which I had the opportunity to chair, several key points were emphasised by the speakers:

a) The emergence of accelerated computing, multimodal systems, and GPU-based computing have enabled the cost-effective development and deployment of large language models, unlocking innovative solutions across multiple domains.

b) The dichotomy between lived and living data has become a focal point as Generative AI expands. Understanding what makes anything unique and exploring creativity through textual means are vital aspects of this evolving landscape.

c) Responsible consideration of pragmatic changes and potential collateral damage is crucial as Generative AI becomes a catalyst for discovering tacit knowledge at extraordinary speed.

d) Generative AI’s capacity to create, discover, summarise, and automate processes presents tremendous opportunities for various industries and domains.

e) Nurturing and advancing the fundamental models that underpin Generative AI requires a collaborative effort from governments and/or companies. Adequate funding and resources are essential to drive research and innovation, pushing the boundaries of what is possible.

Developing its own large language foundational models can be a strategic and advantageous step for India in AI. The reasons supporting this endeavour include the following:

1. Language Diversity: India’s linguistic diversity demands large language models tailored to the Indian context.

2. Cultural Relevance: Models developed in other regions may only partially capture the nuances, cultural references, or contexts specific to India.

3. Data Sovereignty: Building proprietary large language models allows India to maintain control and ownership of its data. This reduces dependence on foreign models, ensuring data sovereignty, privacy, and security.

4. Economic Opportunities: Large language models foster a thriving AI ecosystem within India, driving research and development, creating start-up opportunities, and generating employment.

5. Bridging the Digital Divide: Large language models can bridge the digital divide by enabling access to technology and information for those not proficient in English.

Developing large language models requires substantial resources, expertise, and long-term commitment. Collaboration between academia, industry and government is crucial to creating a sustainable research, development, and deployment framework. Public-private partnerships can play a significant role in driving initiatives to build indigenous large language models.

India should strike a balance between developing localised models and leveraging global expertise. This approach will help achieve the best outcomes in AI while maintaining India’s unique requirements and characteristics. The best place for India to start is to build contextual databases using images, videos, audio, and text. This data should be inclusive of our cities, villages, and tribes; of genders; of our flora and fauna; and our stories and our songs. A key priority is to compile healthcare databases for the sub-continent. Regulations should minimise the role of government and bureaucracy and enable citizens to monetise their data in a transparent, digitally enabled fashion, keeping the corporations open to scrutiny for ethical AI.

With its thriving ecosystem, Bengaluru should take the lead and show the way towards India’s success in Generative AI applications and the development of foundational models.