Amitabh Nag, a seasoned thought leader, serves as the CEO of Digital India Bhashini Division under Digital India Corporation. He holds a distinguished track record of steering Business Units through IT-led transformation programs. With a rich background in esteemed organizations like HP and TCS, Amitabh’s career trajectory has traversed various industries. Furthermore, his extensive engagement in the Government sector, coupled with adept policy-level interactions, has contributed significantly to market-shaping endeavors.
Amitabh’s forte lies in crafting robust Business Strategies and fostering profitable, sustainable ventures by focusing on cost-effectiveness. Notable achievements include overseeing the successful delivery of large-scale projects like Passport Seva and MCA, setting benchmarks in project execution.
As an influential figure within the industry, he actively participates in national-level conferences, serving on committees and sharing insights on IT trends and benefits. Amitabh’s prowess extends to ecosystem development, forging strategic partnerships for Systems Integration, Outsourcing, and collaborating with ISVs and OEMs to pioneer solutions across diverse sectors such as Education, Health, and Smart Cities.
During an exclusive interaction with The Interview World at Startup Mahakumbh 2024, Amitabh Nag, CEO of Digital India’s Bhashini Division, delves into Bhashini, the AI-driven language translation program. He emphasizes its role in eliminating the language barrier in government-to-citizen programs and urges innovative startups to join hands with this initiative to further the nation’s overarching goals. Here are the main takeaways from his interview.
Q: Can you provide insights into the Bhashini program?
A: Bhashini stands as a revolutionary AI-powered language translation tool, serving as a catalyst for collaborative engagement, facilitating research endeavors, and empowering end-users through real-time translation of diverse Indian languages. Its overarching aim is to dismantle the barriers imposed by language diversity, a feat we’ve realized by encompassing 22 languages enshrined in the Indian Constitution. Bhashini’s capabilities extend across multiple mediums, seamlessly translating text, video, audio, and even recognizing optical characters. It comprehensively addresses the spectrum of translation requirements, ensuring a seamless transition between languages. Whether engaging in conversation or crafting written content, Bhashini stands ready to bridge linguistic divides, fostering greater understanding and inclusivity in our interconnected world.
Q: What mechanisms are enabling the government to seamlessly deliver welfare services to citizens through communication channels?
A: Within the Ministry of Agriculture, a significant use case unfolds: the PM Kisan Yojana, streamlining monetary support to farmers. An integral component of this initiative is the PM Kisan eMitra chatbot. This sophisticated platform can interpret voice commands from farmers, ensuring seamless authentication and prompt responses tailored to their specific challenges. Its functionality exemplifies a practical and efficient solution. Moreover, this is just one instance among many others in which technology empowers agricultural communities. These diverse applications demonstrate the breadth of possibilities within this field, showcasing innovative approaches to address the multifaceted needs of farmers and enhance their livelihoods.
Q: What level of accuracy does this system achieve in language translation?
A: The system’s effectiveness hinges on both the domain it operates within and how we apply. In contexts where grammatically correct language is utilized within a specific domain, accuracy can reach impressive levels of 95% or even 100%. However, when venturing into different domains, challenges may arise.
For example, during the launch of a payment application with a voice-based payment app, initial trials with the vanilla model yielded decreased accuracy. Nevertheless, through meticulous data collection and subsequent model training iterations, we managed to elevate performance to a commendable 95% to 100%, thus facilitating the successful deployment of the service. This experience underscores the critical importance of domain-specific training protocols. Every domain boasts its distinct vocabulary, necessitating tailored training approaches to ensure optimal performance and user satisfaction. Therefore, adaptability and continuous refinement are paramount in maximizing the system’s efficacy across diverse domains.
Q: Is Bhashini managed by MeitY or does it rely on a consortium of startups and other organizations?
A: Bhashini, the government-funded program, supported exclusively by the Ministry of Electronics and Information Technology, operates through a structured framework. Various tasks are allocated to individuals by the ministry, each contributing to the program’s objectives. These efforts are then meticulously brought together through aggregation processes. Moreover, research institutes play a crucial role in this initiative by developing the necessary models.
Remarkably, 70 such institutes have actively participated, lending their expertise to the creation of an advanced AI-driven platform. Acting as the central repository for these endeavors, the National Hub of Language Technology effectively consolidates and coordinates the collective contributions, ensuring a cohesive and impactful outcome.
Q: What are the fundamental components and workings of the underlying technology that you could elaborate on?
A: At the heart of our operations lies artificial intelligence (AI), a revolutionary technology reshaping industries worldwide. Our AI model, meticulously honed through the collection and analysis of vast datasets, stands as a testament to innovation. Contrary to common misconceptions, the intricacies of our approach aren’t inaccessible; rather, they represent a strategic utilization of AI’s capabilities. Specifically, we specialize in translation models, a distinct facet within the expansive landscape of AI applications. By prioritizing this niche, we carve a unique path forward, distinguishing ourselves from the broader scope of generative AI and cementing our place at the forefront of technological advancement.
Q: What strategic objectives does the government aim to achieve through this program in the long term?
A: Primarily, we eliminate language barriers across all realms of communication, collaboration, education, innovation, law enforcement, and beyond. Furthermore, our vision underscores the long-term goals of our government. As a nation, India aspires to foster a society where language does not hinder understanding or hinder communication. To achieve this, we harness the potential of artificial intelligence (AI) and other emerging technologies. Through these advancements, we aim to empower individuals to fluently comprehend and communicate in any language. This promotes inclusivity, cooperation, and progress on a global scale.
Q: What advice would you offer to startups interested in developing solutions for Bhashini?
A: We’re currently collaborating with six or seven dynamic startups on a focused initiative. These startups have united their efforts to address a pressing need – the inadequacy of previous language translation solutions, which constrained people’s options. Fortunately, significant progress has been made, with a myriad of effective tools now available, including video transcription capabilities.
In our quest to further advance in this domain, we eagerly invite innovative startups to join forces with us. Together, we can harness the vast potential of Indian language technology solutions, applications, datasets, and AI models on a scale previously unseen. Startups should contribute their ingenuity by developing pioneering multilingual applications and services. Additionally, we call upon the broader industry to play a pivotal role by fostering the development of open-source software, offering robust storage solutions, and generously contributing computational resources for model training. Therefore, through collective effort and collaboration, we can usher in a new era of language technology innovation, benefiting countless individuals and industries alike.
