JD.com and Former OpenAI CTO Mira Murati Have Bet on the Same AI Track
JD.com and Mira Murati's Thinking Machines Lab are converging on the same AI frontier: proactive visual-language interaction models. JD just open-sourced JoyAI-VL-Interaction, the world's first full-stack open-source model of its kind. Unlike traditional "turn-based" AI that waits for user prompts, this model actively analyzes continuous video streams, autonomously deciding when to respond, stay silent, or delegate complex tasks. It prioritizes vision as the primary driver for decision-making in physical-world scenarios like elderly fall detection, live sports commentary, or warehouse monitoring.
The 8-billion-parameter model is designed for practical deployment, running on a single RTX 3090 GPU with sub-second latency. Its "full-stack" open-source release includes the model, inference system, and dataset, aiming to catalyze a developer ecosystem. JD's strategy is underpinned by its vast operational footprint in retail, logistics, and healthcare, which provides crucial real-world data for training. The move signals a broader shift in AI competition from screen-based Q&A to active participation in the physical world.
marsbit06/24 10:23
