IT companies are increasingly focusing on small language models (SLMs) designed for specific industries, driven by rising hardware costs and a demand for cost-effective solutions. Executives from firms such as Infosys and HCLTech emphasize the benefits of these models in improving operational efficiency and enabling tailored applications, signaling a shift towards more specialized AI technologies in the market.
In the realm of Generative AI, IT firms are increasingly investing in small language models (SLMs) designed for specific industries such as banking, IT operations, cybersecurity, and broader enterprise applications. Salil Parekh, CEO of Infosys, highlighted their development of four distinct SLMs tailored to these sectors, emphasizing their potential in streamlining operations.
C. Vijayakumar, CEO of HCLTech, noted that the cost efficiency of utilizing smaller, highly specific models significantly enhances their appeal. He stated, “I think the real difference is the cost at which you can do this, is going to make a big difference because Agentic uses small or very small models and very, very specific models. That makes the whole price performance of what we want to achieve much more compelling.”
Vijayakumar anticipates an accelerated adoption of SLMs, asserting that they positively impact legacy modernization programs driven by Generative AI. The push for these targeted models is fueled by rising hardware costs, as noted by Abhigyan Malik from Everest Group. He remarked, “With hardware prices shooting through the roof throughout 2024, the need for optimized versions tailored for a specific use-case was becoming the need of the hour,” underscoring the importance of cost-effective solutions for enterprises.
Moreover, the cost of leveraging large language models (LLMs) and conversational AI has seen a drastic decrease of over 85% since early 2023, according to Vijayakumar. Tech Mahindra has shifted its focus from LLMs to SLMs and compact language models, finding significant relevance in addressing specific customer needs without incurring excessive computational resources. Mohit Joshi, CEO of Tech Mahindra, mentioned, “Customers really find the relevance of these models in small use cases. They allow customers to solve fairly specific problems without using up too much of compute or carbon.”
Infosys’s Parekh explained that these models utilize proprietary client data in combination with standardized industry data, enabling tailored solutions for clients. This approach assists companies in building specialized models that meet their distinct operational requirements, thereby enhancing efficiency and cost-effectiveness across various sectors.
The trend towards developing small language models in IT is gaining momentum, driven by the need for cost-effective, specific solutions tailored to various industries. As companies like Infosys, HCLTech, and Tech Mahindra adapt their strategies to focus on SLMs, the significant reduction in costs for these technologies promises to accelerate AI adoption and provide firms with a competitive edge. The insights from industry leaders reveal a promising future for targeted AI applications across diverse business sectors.
Original Source: m.economictimes.com
Leave a Reply