Yasmeen Ahmad, who is the managing director of strategy and outbound product management for data, analytics, and AI at Google Cloud, recently shared insights at VB Transform about large language models (LLMs) and the future of AI. She highlighted that while bigger models do improve with size, they can be outperformed by smaller models trained on domain-specific information. This emphasizes the importance of domain-specific industry data in giving models power.
Ahmad also discussed the significance of fine-tuning and retrieval augmented generation (RAG) techniques in training models on enterprise domains. Fine-tuning helps LLMs learn the language of a specific business, while RAG enables real-time connections to data sources like documents and databases. These techniques are crucial for applications such as financial and risk analytics.
Moreover, she emphasized the value of LLMs’ multimodal capabilities, which allow them to work with various types of data such as text, images, and videos. Ahmad mentioned that Google’s study showed a significant improvement in customer experience when multimodal data was utilized. This underscores the importance of LLMs in understanding the complexity of organizations by accessing diverse data sources.
In addition, Ahmad highlighted the importance of conversational AI in LLMs. She mentioned that enabling question-answer interactions and providing semantic context and metadata are essential for obtaining specific and accurate answers. This conversational aspect is crucial for the next generation of AI, moving beyond isolated interactions to more interactive and engaging experiences.
Furthermore, Ahmad discussed how LLMs are evolving to mimic human brain functions, such as breaking tasks into sub-tasks, understanding cause and effect, and learning honesty. These advancements are leading to faster real-time capabilities and spawning new opportunities for businesses.
Overall, Ahmad’s insights shed light on the evolving landscape of AI, emphasizing the need for domain-specific data, multimodal capabilities, conversational AI, and the potential for LLMs to drive innovation in various industries. As organizations continue to leverage AI technologies, understanding these key learnings will be essential for unlocking the full potential of AI in business applications.
