Nvidia’s DLSS 5 uses generative AI to boost photo-realism in video games, with ambitions beyond gaming

During the recent Nvidia GTC keynote, CEO Jensen Huang unveiled DLSS 5, the latest advancement in the company's AI graphics technology, aimed at enhancing realism in video games while optimizing computing efficiency. This innovative system integrates conventional 3D graphics with generative AI models that can intelligently predict and complete image elements. The result is a capability that allows Nvidia's GPUs to render intricate scenes and lifelike characters without the need to generate every detail from the ground up. "We have merged controllable 3D graphics, the solid foundation of virtual environments, with generative AI and probabilistic computing," Huang explained during his presentation. He emphasized that this fusion of structured data and generative AI empowers developers to craft content that is not only stunning and impressive but also manageable. Huang forecasted that this integration of structured information and generative AI will proliferate across multiple industries. He underscored the importance of structured data as a cornerstone for reliable AI. While gaming currently represents a smaller fraction of Nvidia’s revenue compared to its historical significance, Huang highlighted the potential of DLSS 5 as a model for a wider computing evolution that could transcend the gaming sector and impact enterprise applications. He cited enterprise platforms such as Snowflake, Databricks, and BigQuery as examples of structured datasets ripe for future AI analysis and insight generation. "In the future, these structured data frameworks will be leveraged by AI, operating at speeds far surpassing our capabilities," Huang asserted. "Emerging AI agents will utilize both structured and unstructured databases, reflecting a significant portion of our world’s information landscape."