Nvidia's GTC conference keynote on Monday centered on three blockbuster reveals from the AI chip giant, though not all of them received a warm reception from everyone in the industry.
An AI-upscaling technology, called DLSS 5, which gives video games a more cinematic experience, faced immediate backlash from gamers who accused it of dramatically altering characters' faces, with some even comparing it to AI slop.
Another announcement was an open-source AI agent stack called NemoClaw, capable of running claws, or autonomous AI agents. CEO Jensen Huang highlighted a focus on its data center platform, Vera Rubin, and the trend toward agentic AI. He introduced a new, optimized processor called the Vera CPU, designed to handle agentic AI systems, physical AI and more.
If you missed the keynote on Monday, here are the three big announcements you need to know about.
DLSS 5: bringing AI-powered visual fidelity to games
Nvidia promised impressive advancements in computer graphics, including enhanced performance and image quality, through AI-powered upscaling to create highly realistic videos that look like a film.
According to Nvidia, DLSS 5 brings real-time neural rendering that "infuses pixels with photo-real lighting and materials." The company says the AI model is trained to understand features such as characters, fabrics, translucent skin and environmental lighting systems -- all from a single frame. DLSS 5 then generates upscaled visuals for the scene.
But the unveiling of DLSS 5 drew immediate scrutiny from gamers. "Nvidia's DLSS 5.0 footage showed the tech dramatically changing how beloved characters look on the fly, seemingly without input from the game's creators, which rubbed gamers the wrong way," said CNET Managing Editor David Lumb.
Lumb noted that the infiltration of generative AI has generally received a chilly reception among gamers: "When they've gotten wind that a game has used the technology, they've turned on it pretty quickly."
CEO Huang dismissed criticisms of DLSS 5, saying developers can fine-tune the model to match their creative vision rather than submitting to AI.
... continue reading