Synopsis: In this Techstrong AI video interview, SymphonyAI CTO Raj Shukla dives into the impact that all the data being collected to drive generative artificial intelligence (AI) will have on IT infrastructure.
In this Techstrong AI interview, host Mike Vizard speaks with Raj Shukla, CTO of SymphonyAI, about the complexities enterprises face when integrating AI into their data infrastructure. Shukla explains that while large language models (LLMs) trained on public data can provide immediate responses, applying AI within enterprises requires meticulous data organization and preparation. Many companies underestimated the effort needed to prepare data for AI, particularly when dealing with unstructured data like documents and images. Shukla also discusses how unsupervised and semi-supervised AI techniques can add value without extensive human annotation, highlighting how advances in tools like LLMs can automate much of this work.
As the conversation progresses, Shukla addresses the importance of understanding data workflows and orchestration layers in AI implementation. He describes how Symphony AI employs a layered approach to handle real-time processes, such as financial crime detection, and how companies need to evaluate the complexity of their AI workflows before choosing between a build-versus-buy strategy. Shukla advises organizations to classify their use cases as generic, industry-specific or company-specific to make better-informed decisions. He reflects on the rapid evolution in AI, emphasizing how the journey is one of constant iteration, with new advancements allowing enterprises to adapt more quickly than ever.
