In this episode of FO° Talks Predata CEO Sanjay Mittal joins Fair Observer Chief Strategy Officer Peter Isackson to dive into what they call the DeepSeek revolution in AI. They argued that DeepSeek — an open-source model built on Meta’s Llama 3 —  dominant proprietary systems from companies like OpenAI and Google. By achieving comparable performance with fewer resources, DeepSeek shows how necessity can drive smart, efficient innovation. Unlike closed models, DeepSeek invites scrutiny, thanks to its transparent design.

Mittal and Isackson highlighted how open-source AI can democratize the field. Developers and researchers can collaborate more freely, build faster, and better understand how these systems work—advantages that secretive models can't offer.

They also explored “distillation,” a process that creates smaller, faster models fine-tuned for specific tasks. When paired with AI agents—tools that act autonomously—these streamlined models can power more practical and accessible real-world applications.

To close, they tackled deeper questions: Does AI truly “think,” or does it just mimic patterns? How should we interpret its growing efficiency and capability? And could we eventually build artificial general intelligence—machines that think and learn like humans? While they didn’t claim to have final answers, they made one point clear: the future of AI may depend as much on openness and shared effort as on raw computing power.

The views expressed in this article/video are the author’s own and do not necessarily reflect Fair Observer’s editorial policy.