CEOInterviews.AI
Start App
Try: AI strategy tariffs earnings guidance layoffs interest rates growth

Ben Mildenhall

Co-founder, World Labs

Search every verified Ben Mildenhall interview, podcast appearance, and on-the-record quote — each transcript cross-checked by AI and human review to confirm speaker identity. Ben Mildenhall, co-founder of World Labs, discussed the company's Atlas world model during a September 2026 appearance on an a16z podcast with fellow co-founders Fei-Fei Li and Justin Johnson. He described Atlas as built on "new view prediction," a primitive he said generates spatially contextualized pixels by predicting what a scene looks like from different positions in space and time, in contrast to next-token prediction in large language models or next-frame prediction in video models. He stated that this approach combines generation and 3D reconstruction into a single model and that the team believes it is a fundamental step toward spatial intelligence. Mildenhall also addressed the model's 3D consistency, attributing it partly to a belief in the scaling hypothesis, while noting that exact architecture choices and data mixtures are where challenges lie. He cited a practical example from the film The Matrix, saying Atlas could recreate a similar multi-camera effect with just three cameras, without studio capture or green screens. He expressed confidence in the approach, stating that the model improved significantly with increased size and training duration.

Recent appearances

  • Why Fei-Fei Li Is Betting on Spatial Intelligence

    World Labs co-founders Fei-Fei Li, Justin Johnson, and Ben Mildenhall join a16z General Partner Martin Casado to discuss Atlas, their latest world model, and what it reveals about the pursuit of spatial intelligence. At the center of Atlas is what the team calls “new view prediction”: given images or views of a scene, the model predicts what that environment should look like from a different position in space and time. This brings generation and 3D reconstruction into the same model, and raises a broader question about whether predicting views could become a useful primitive for understanding…

    Watch on YouTube ↗

  • L13b Neural Radiance Fields -- Guest Lecturer Ben Mildenhall

    CS294-158 Deep Unsupervised Learning UC Berkeley Spring 2024 Instructors: Pieter Abbeel, Kevin Frans, Philipp Wu, Wilson Yan course website: https://sites.google.com/view/berkele...

    Watch on YouTube ↗

  • 3DV 2024 Keynote - Ben Mildenhall - 21.03.2024

    ... Ben obtained his PhD in computer science from UC Berkeley in 2020 uh from 2021 to until recently he was a research scientist ...

    Watch on YouTube ↗

Other executives at World Labs

The summary and quote tags on this profile are produced with AI assistance from verified, first-person interview transcripts, then checked by our team to confirm the speaker's identity and the accuracy of every quote. See how we verify →