Anthropic has proven: The AI Claude controls a robot dog tens of times faster than humans
The artificial intelligence industry has made another breakthrough, and this time it has gone beyond simple text generation. Anthropic has unveiled updated results from the Project Fetch experiment, demonstrating that language models are capable not only of writing code but also of fully controlling physical robots. My analysis shows this is a turning point for all of robotics.
During the second phase of testing, the Claude Opus 4.7 model took full control of a four-legged robot — a robot dog. The results are staggering: the neural network completed configuration and control tasks 20 times faster than teams of human engineers working under similar conditions. Compared to last year's figures, Opus 4.7 proved to be 18 times more productive than teams using older AI versions and 37 times faster than humans working without chatbot assistance.
How AI Took Control
Unlike the first phase of the experiment (August 2024), where AI acted only as an assistant to humans, the model now worked almost autonomously under minimal researcher supervision. Claude Opus 4.7 independently completed a whole range of tasks:
- connected to video sensors and LiDAR;
- wrote a program for manual robot control;
- created a path monitoring system;
- configured an object recognition algorithm.
The quality of the code generated by the neural network deserves special attention. Its volume turned out to be 10 times smaller than that of human teams, while being more efficient and completing tasks faster. This confirms my thesis: AI not only automates processes but also optimizes them at a fundamentally new level.
Limitations and Prospects
However, one should not think that AI has already completely replaced humans in physical robotics. Despite success in navigation and programming, Claude experienced serious difficulties with precise physical actions. The model managed to guide the robot to the target but failed at the task of gently pushing a ball to the right spot. This requires complex real-time feedback, in which humans still outperform AI.
The experiment's authors emphasize that progress in robotics has been a side effect of the general scaling of language models. Anthropic did not introduce specialized algorithms for controlling hardware — this happened naturally during the development of basic architectures. I believe the industry is truly entering an era of "physical AI agents," where neural networks will be able to use standard tools and equipment as effectively as they currently work with software code.
My analysis: The Anthropic experiment is not just a technology demonstration but a serious signal for investors and developers. We are witnessing a convergence of language models and robotics that will open up new markets. However, the key challenge will remain integrating AI into real physical systems with feedback — this is where human experience is still irreplaceable.