No Priors Ep. 37 | With Kawal Gandhi

Model Training

Kawal Gandhi discusses the intricacies of training generative AI models, emphasizing the importance of understanding organizational maturity and strategic alignment. He advises enterprises to consider efficiency, productivity, and creativity when implementing AI, highlighting the need for trust in AI systems to ensure they function as intended 1. Gandhi also notes the growing interest in multimodal AI, which combines text, images, and audio, and stresses the importance of safety and cost management in these applications 2.

How can I make my workflow better so I can invest back into my group? And then how can I make productivity better? Then I can make them creative.

--- Kawal Gandhi

He sees significant potential for AI in industries like gaming, where multimodal applications are already being explored.

   

Inference Scaling

Scaling AI inference presents unique challenges, particularly in the context of hardware limitations like the Nvidia GPU shortage. Gandhi highlights the importance of providing customers with options and flexibility in their AI infrastructure, ensuring that they can scale applications efficiently without being hindered by hardware constraints 3. He emphasizes Google's commitment to making AI platforms accessible and adaptable, allowing customers to leverage existing investments while accommodating new models and technologies 4.

We think about going back that the customer should have availability and optionality.

--- Kawal Gandhi

This approach helps organizations focus on deploying AI solutions effectively, even amidst supply chain challenges.