Can generative AI become marketers' secret weapon for regional language content?

New Delhi: India, a nation of over 1.4 billion people, is a marketer’s dream. Its diverse population, with a myriad of languages, cultures, and preferences, presents a vast, untapped market. However, the country’s linguistic complexity is also a significant challenge for businesses aiming to leverage the power of Generative Artificial Intelligence (GenAI).

India boasts 22 official languages and thousands of dialects, creating a complex linguistic landscape. This diversity is a double-edged sword. On one hand, it represents a vast, untapped market for AI-powered solutions. On the other hand, developing accurate language models for such a diverse population is a daunting task.

While GenAI has made remarkable strides in recent years, its ability to cater to India’s linguistic diversity remains a significant hurdle. Despite advancements in natural language processing (NLP), many regional languages still lack adequate data and computational resources. This gap translates to missed opportunities for businesses across various sectors.

• Personalised marketing: GenAI excels at creating personalised content and experiences. However, without robust language models for regional languages, marketers struggle to deliver tailored messages to a significant portion of the Indian population. 

• Customer service: AI-powered chatbots and virtual assistants can revolutionise customer service. However, their effectiveness is hindered by language limitations. A significant portion of the Indian population still prefers to interact in their native language, and a lack of language support can lead to frustration and customer churn.

• Market research: Understanding consumer sentiment and preferences is crucial for effective marketing. GenAI can analyse vast amounts of data to uncover valuable insights. However, when a significant portion of the data is in regional languages, the analysis becomes challenging, limiting the depth of market research.

• Content creation: GenAI can generate high-quality content at scale. However, without language proficiency, it struggles to create culturally relevant and engaging content for regional audiences. This limits the ability of businesses to reach and connect with consumers in their preferred language.

Garima Saxena, Senior Research Associate at The Dialogue, highlighted an example: “Platforms like Bhashini, India’s national language database, are paving the way for multi-model generative AI services that cater to non-English speaking populations. This development is crucial for bridging the digital divide and making AI-powered marketing more inclusive.” 

She further said that the development of purpose-driven, narrow AI models for specific use cases will be key to overcoming these challenges and improving the effectiveness of AI in regional marketing.

However, data scarcity is a primary hurdle. While there's a wealth of digital content in English and a few major Indian languages, many regional languages have limited data availability. This dearth of data hinders model training and accuracy. Furthermore, the intricate grammar and scripts of some Indian languages add to the challenge.

“The challenge with GenAI in India lies in the sheer diversity of languages,” said Manoj Karunakaran, Tech Head, BC Web Wise. “While translation between major languages has improved significantly, the nuances of regional dialects and the lack of quality data remain significant obstacles.”

Despite the challenges, India is witnessing a GenAI revolution. Startups, tech giants, and research institutions are heavily invested in developing language models for Indian languages. Government initiatives are also playing a crucial role.

Data augmentation techniques are being employed to expand limited datasets. Multilingual models, capable of handling multiple languages simultaneously, are showing promise. A growing pool of linguists and computer scientists is contributing to the development of robust language technologies.

“The ability to communicate with customers in their native language is a game-changer. Whether it's e-commerce, customer service, or any digital platform, the use of local languages is becoming increasingly important,” said Karunakaran.

Rohan Shah, Co-Founder at Logicloop and Realatte Ventures LLP, highlighted the potential of AI in translation and video content creation. “AI can be immensely helpful in translating marketing materials into different languages, enabling businesses to reach a wider audience. The diversity of India presents a massive opportunity for growth," said Shah.

Shah also emphasised the increasing role of AI in video content creation. “AI-generated videos are becoming more sophisticated and will likely become a mainstream tool for digital marketers in the next few years,” he added.

However, both experts agree on the importance of human oversight. “While AI can automate many tasks, human intervention is crucial to ensuring the accuracy, relevance, and cultural sensitivity of content,” said Karunakaran.

While still in its early stages, there are promising examples of GenAI tools capable of generating content in Indian languages. Discussing the relevance of GenAI’s conversational ability at Microsoft’s Cafe Copilot, Bhaskar Basu, Country Head, Microsoft India, mentioned ‘hinglish’ as being a common approach for many users. He went on to present several examples of Copilot being capable of understanding and generating Hinglish outputs, demonstrating the potential for more complex language interactions. 

This shows how there is a significant step towards bridging the gap between English-centric development and the vast Indian user base. Some startups and research institutions are developing AI models tailored for specific Indian languages, showing promising results in text generation, translation, and summarisation.

Meanwhile, the limitations of GenAI have led to a surge in alternative content formats that cater to regional audiences:

As Saxena said, “GenAI can be beneficial across various forms and mediums of marketing. It excels in content creation and generates diverse materials in multiple languages and formats. As demonstrated with UPI payments, voice-driven interfaces can make services more accessible. Combining text, voice, and visual elements, multi-modal marketing becomes more feasible with AI.”

There is growing optimism about the future of GenAI in India. Several initiatives are underway to address the language gap:

“AI ad targeting in India walks a tightrope,” said Bhavik Mehta, CEO of Thinkin' Birds Communications. Mehta emphasised the importance of a multi-pronged approach, including A/B testing and diverse content partnerships. “We can harness AI's personalisation power across languages while incorporating elements that spark discovery.”

The inability of GenAI to fully tap into India's linguistic diversity has significant economic consequences. A report by NASSCOM estimates the potential market size for language technology in India to be $8 billion by 2025. However, this potential remains unrealised.

Moreover, the lack of AI-powered tools for regional languages hinders the growth of SMEs. These businesses often rely on limited marketing strategies, hindering their ability to scale. AI-powered solutions could help them reach a wider audience and compete more effectively.

“The accuracy of GenAI in understanding and responding to regional dialects is still a major challenge. To truly harness the potential of AI for India, we need to invest in developing robust language models that can handle the complexities of our linguistic landscape,” said Karunakaran.

Building a strong foundation herein becomes a necessity. As a part of her research, Garima believes the following could be the beginning towards the industry’s preparedness for GenAI effectiveness in marketing:

• Professionals will need to constantly update their skills to keep pace with evolving AI technologies.

• Understanding the ethical implications of AI use in creative work will be crucial.

• Learning to collaborate with AI tools effectively will be a key skill.

• Educational programs that combine creative and technical disciplines will be vital in nurturing future creative technologists.

• The ability to pivot between creative and technical thinking will be essential.

Looking ahead, the creative industry should embrace GenAI’s loopholes as a possible way to balance both modern and traditional marketing, while anticipating further development in the field as below:

• The contentious issue of copyright and intellectual property rights is at the forefront of these discussions in the creative industry. The crux of the matter lies in whether it's permissible for companies to use copyrighted works to train AI algorithms under existing ‘fair use’ clauses. This practice has put AI companies in direct conflict with artists and rights holders, who argue that such use cannot be considered fair when it directly competes with their work in the same marketplace.

• Another significant concern is the call for transparency in AI-generated content. There's a growing demand for clear attribution and disclosure when AI is involved in creative works, similar to product labelling in other industries. However, this presents complex challenges in determining the boundaries between human and AI contributions, especially given the intricate integration of AI in various stages of the creative process. 

• Ethical considerations are becoming increasingly prominent in the field. Some AI technologists are adopting pro-artist stances, recognising the need to respect creators' rights. Initiatives like the Human Artistry Campaign are emerging to advocate for policies that protect artists' interests in this evolving landscape.