Discovery Loop
Points and comments are a snapshot, not live.
Discovery Loop automates experimental loops with AI to accelerate scientific and engineering discovery.
The startup, founded by Jeff Dean, Sanjay Ghemawat, Quoc Le, and Oriol Vinyals, aims to automate the entire experimental loop using frontier AI models and large-scale infrastructure. It will initially focus on machine learning research and engineering, acting as its own first customer. The team's past work includes Google Search, MapReduce, TensorFlow, TPUs, AlphaFold, and Gemini. Their long-term goal is to tackle National Academy of Engineering Grand Challenges such as engineering better medicines and making solar energy economical.
What commenters are saying
The thread is split between skepticism and excitement. One camp criticizes Jeff Dean's track record, listing failures like LevelDB's crash issues, TensorFlow losing to PyTorch, and Gemini's fake demo. Another camp dismisses these as sour grapes. Many commenters note the website's Claude-generated aesthetic, calling it bland. Others debate whether LLMs are plateauing, with some arguing self-evolving agents are the next frontier. Several commenters view the startup as a lifestyle business or hobby rather than a high-growth venture, given its public benefit corp structure.