Using localised AI solutions for business growth in South-east Asia

FOR global companies expanding to or operating in South-east Asia, integrating artificial intelligence (AI) chatbots at scale poses a unique challenge. After all, the region is home to more than 1,000 languages.

An AI model trained on standard Western-based large language models (LLMs) is likely to misinterpret a prompt or produce gibberish responses when responding to customers in Malay, Thai or Bahasa Indonesia.

This translates to poor customer experiences, undermining efforts to provide localised services in one of the world’s fastest-growing markets.

Recognising the region’s linguistic diversity, Nvidia collaborated with AI Singapore to develop SEA-LION, a first-of-its-kind language model trained on 11 South-east Asian languages.

Nvidia supported the fine-tuning of the model and introduced the first Nvidia NIM inference microservice for the region. NIM microservices simplify the building and deployment of AI models for businesses, working like Lego blocks chosen from a catalogue that developers can use to assemble specialised pieces – such as speech, reasoning and vision – to create customised AI systems quickly.

This is just one of the initiatives that Nvidia, the world leader in accelerated computing, has rolled out to help businesses create more localised and accurate AI applications.

“As one of the fastest-growing digital economies, South-east Asia is a powerhouse for AI development and adoption. By making it easier for developers to build sovereign and regional language models, Nvidia is helping to ensure that everyone can participate in and benefit from the digital revolution,” says Dennis Ang, senior director of enterprise, Nvidia.

Watch this video to see how global companies are unlocking AI innovation with Nvidia:

Supporting companies’ AI transformation journey

Having captivated business leaders across industries, generative AI has the potential to further catalyse South-east Asia’s digital economy, which could grow from $300 billion to nearly $1 trillion by 2030, according to the e-Conomy SEA 2023 report. 

Companies like OCBC Singapore have already developed generative AI applications on Nvidia-optimised infrastructure that help employees with coding, internal FAQs and customer service. 

In Vietnam, tech company VinBrain deployed AI models in more than 100 hospitals to support doctors by automating tasks, including detecting abnormalities in chest X-rays, using Nvidia’s MONAI (Medical Open Network for AI) healthcare platform.

However, technology and talent gaps exist in the region. Challenges to developing AI applications at scale include a lack of technical expertise, data privacy and security concerns, and the need for substantial infrastructure investments.

Nvidia offers comprehensive support to companies looking to integrate generative AI into their business processes. 

Says Ang: “Our technology provides the computational power needed for complex applications, assisting the finance, healthcare and retail industries in realising the full potential of generative AI.”

To bridge the talent gap, Nvidia offers professional certifications for foundational and advanced AI through its Deep Learning Institute. The programme provides valuable resources and hands-on training in AI, accelerated computing and data science.

“Engaging in high-quality professional development for AI should be integral to any AI transformation initiative,” Ang adds.

Quantifying AI investments, from pilot to profit

Business leaders may hesitate to invest in generative AI without a clear return on investment (ROI). Ang advises companies to start with small, manageable projects to demonstrate AI’s value to stakeholders while minimising risk.

He recommends identifying specific areas where AI can make a tangible impact on the company’s strategic goals, such as automating routine tasks to enhance operational efficiency or improving customer service through AI-powered chatbots. This approach helps businesses avoid the classic pitfall of adopting technology without a clear use case.     

“By focusing on particular uses, businesses can quickly see the benefits without immediately committing to major changes,” says Ang.

To help businesses quantify the value of generative AI, Nvidia provides benchmarks that highlight improvements in speed and efficiency of its AI solutions.

Its product suite includes ROI analysis tools and frameworks to measure productivity improvements, cost savings and revenue growth. The company also provides consultative support, working closely with clients to tailor AI solutions to their needs to measure productivity improvements, cost savings and revenue growth. The company also provides consultative support, working closely with clients to tailor AI solutions to their needs.

“This hands-on approach ensures enterprises can design, deploy and optimise their AI projects effectively,” says Ang.

Guarding against data leaks and AI misuse

Other issues that business leaders are concerned about include accidental data leaks and the potential misuse of AI systems.

LLMs may inadvertently expose sensitive data due to training issues or improper handling. Unless explicitly prompted, AI-generated code might contain security flaws that hackers could exploit. 

To address these concerns, Ang suggests companies establish multiple safeguards and oversight mechanisms to ensure AI systems adhere to rules and company values. 

Nvidia offers tools like NeMo Guardrails, which lets companies create custom rules to control AI behaviour in real-world situations. Instead of fully automating AI outputs, having employees review and correct AI processes can further eliminate inconsistencies. 

“AI is a journey, not a destination,” says Ang. “Businesses need to continuously monitor and refine their AI applications to adapt to evolving needs and technological advancements.” 

“Taking a measured approach can help business leaders overcome hesitation and harness AI’s transformative power for their strategic decision-making.”

Learn more about how Nvidia can transform your company’s AI journey.