Nvidia CEO Jensen Huang used the company’s keynote at Nvidia GTC on Monday to unveil DLSS 5, a new version of its AI-based graphics technology aimed at improving visual realism in video games while reducing compute requirements.
Combining Graphics and Generative AI
The DLSS 5 system integrates traditional 3D graphics data with generative AI models. This approach allows Nvidia’s GPUs to predict and reconstruct parts of an image, enabling the creation of detailed environments and lifelike characters without rendering every element directly.
Huang explained that the system merges structured 3D graphics, which represent the ground truth of virtual environments, with generative AI, which relies on probabilistic computing. He noted that one component is fully predictive, while the other produces results that are less deterministic but visually realistic.
Huang said this combination of structured data and generative AI gives developers greater control while maintaining high visual quality. He added that this model of computing could extend beyond gaming into other industries.
According to Huang, structured data will play a central role in building reliable AI systems. He described it as the foundation for future AI applications that can generate insights while maintaining accuracy.
Huang pointed to enterprise platforms such as Snowflake, Databricks, and BigQuery as examples of structured data systems that could support future AI models. He said upcoming AI agents will rely on both structured and unstructured data sources to operate more efficiently.
He added that these systems will process information faster than humans, using structured databases alongside generative data systems that represent a large portion of global information.
This article is so bad. This has been universally mocked.
Afaq you need help.