A Malaysian startup, Mesolitica, has made significant strides in artificial intelligence by developing a generative AI large language model (LLM) specifically tailored for the Malaysian language. This innovative model, which is designed to enhance customer service for small and medium enterprises (SMEs), was created in partnership with Amazon Web Services (AWS), a leading player in cloud technology.
Launched to improve engagement with local customers, the model leverages the extensive cloud capabilities of AWS, allowing Mesolitica to address the specific linguistic and cultural nuances found in Malaysia. The generative AI system has been trained to manage a variety of regional dialects, slang, and colloquialisms, encompassing Bahasa Malayu and 16 other local languages. This training aims to resolve the shortcomings faced by businesses that rely on AI systems pre-trained in English and predominantly Western contexts, such as issues of cultural ignorance and communication inaccuracies.
Dr. Kev Lim, the CEO and Founder of the health-tech startup Qmed Asia, highlighted the model's impact on healthcare operations. Speaking to PaySpace Magazine, he remarked on the challenges faced in understanding the diverse local languages used by Malaysian patients. He shared that by implementing Mesolitica’s MaLLam model, his organization improved its capabilities in deciphering local speech patterns, which subsequently enhanced the accuracy of medical note-taking, improved patient communication, and allowed for the delivery of higher-quality healthcare services.
The implications of this innovation extend beyond the private sector. The Malaysian government is reportedly considering the integration of the MaLLaM AI system into its operations, aligning it with national objectives such as AI sovereignty and enhanced local data management. Government-powered AI assistants could potentially provide citizens with quick and precise information across multiple languages and regional dialects, further stimulating the efficiency of data processing and engagement in Malaysia’s diverse landscape.
In the educational context, MaLLaM presents opportunities for language learning and research, thus aiding in the study and preservation of Malaysia’s cultural heritage through its myriad languages and dialects.
Khalil Nooh, co-founder and CEO of Mesolitica, pointed out the strategic choice of AWS as a partner to optimise their project’s outcomes. He noted that AWS has enabled substantial savings in computing costs, along with an increase in throughput, which translates to heightened operational efficiency. "We managed to significantly accelerate our ability to deploy proofs-of-concept by using AWS," Nooh explained, highlighting the advantages of AWS’s cost-effective AI computing resources and advanced machine learning tools.
Mesolitica has been harnessing several AWS services, including Amazon EC2, Amazon EKS, and Amazon SageMaker, to facilitate efficient training, deployment, and management of its AI models. The infrastructure provided by AWS not only reduces costs through GPU-accelerated capabilities but also comes with the added benefits of participating in the AWS APJ Generative AI Spotlight accelerator program.
As the landscape of AI and automation continues to evolve, Mesolitica's initiatives represent a growing trend towards the localisation of technology, ensuring that businesses in Malaysia can operate effectively within their unique cultural and linguistic frameworks.
Source: Noah Wire Services