With the emergence of OpenAI and ChatGPT, artificial intelligence remains a primary focus for businesses seeking to enhance value propositions. Large language models play a pivotal role in catering to industry-specific requirements, preferences, and tastes, as highlighted by Muddu Sudhakar, the co-founder and CEO of Aisera Inc.
Sudhakar foresees the next wave of value residing in Domain-Specific Language Models (DSLMs) rather than the foundational models. These DSLMs, also referred to as Small Language Models (SLMs), are tailored to specific domains such as IT, HR, and finance, with each enterprise having its unique model.
In a discussion with industry analyst John Furrier at the “Supercloud 5: The Battle for AI Supremacy” event on theCUBE, Sudhakar emphasized that DSLMs will adopt a platform-as-a-service (PaaS) approach, acting as the bridge between applications and foundational layers. Securing these DSLMs is crucial as they serve as the internet protocols for enterprises.
As technology advances and GPUs evolve to enhance performance optimization in alignment with smarter chips, the demand for the next generation of GPUs to accelerate DSLMs becomes imperative. Aisera is positioned to leverage this technological progression to drive faster DSLMs.
The impact of generative AI on both consumers and enterprises is reshaping the data management landscape. Sudhakar stressed the importance of data ownership, highlighting the need for domain-specific data to remain within the control of businesses while ensuring data privacy and permission protocols are respected.
The ongoing debate between open-source and closed-source AI models persists, with Sudhakar advocating for a balanced approach that leverages the strengths of both paradigms for optimal value creation. This nuanced strategy is essential for leveraging the capabilities of foundational models effectively.
