The AI model Claude Opus 4.7 has surpassed humans in controlling a robot dog: speed is 20 times higher
The Anthropic Institute has unveiled updated results from the Project Fetch experiment, demonstrating rapid progress in integrating artificial intelligence with physical robotic systems. In the new testing phase, the Claude Opus 4.7 model completed the full cycle of setting up and controlling a robot dog 20 times faster than teams of experienced human engineers.
Autonomy Without Specialized Algorithms
Unlike the first phase of the experiment, which took place in August 2024, when AI only assisted employees with no experience in robotics, the current version of Claude worked almost autonomously. Under minimal researcher supervision, the neural network independently performed a number of critically important operations:
- connected to video sensors and LiDAR;
- wrote a program for manual robot control;
- created a trajectory monitoring system;
- configured an object recognition algorithm.
The key metric is speed. Claude Opus 4.7 proved to be 18 times faster than a team using previous AI versions and 37 times faster than humans working without chatbot assistance. Moreover, the code generated by the neural network turned out to be 10 times more compact than code written by human teams, indicating higher efficiency and cleaner architecture.
Progress as a Side Effect of Scaling
It is important to emphasize that this breakthrough is not the result of implementing specialized algorithms for controlling hardware. As the developers note, progress in robotics has become a side effect of the general scaling of language models. This confirms the thesis that fundamental improvements in large language models can yield unexpected but extremely valuable applied results.
However, there were limitations. Despite impressive performance, Claude still faces serious difficulties with precise physical actions. The model successfully guided the robot to its target but could not accurately push a ball to the desired point. This requires complex real-time feedback, where humans still maintain an advantage.
The Era of Physical AI Agents
Anthropic believes the industry is on the verge of an era of "physical AI agents." In the near future, neural networks will be able to use standard tools and equipment as effectively as they currently work with software code.
Recall that on June 13, Anthropic was forced to halt access to the Fable 5 and Mythos 5 models due to a directive from the U.S. government under export controls. This once again underscores how rapidly the regulatory environment around advanced AI systems is changing.
My analysis: The results of Project Fetch are not just a demonstration of the capabilities of another model. It is a signal that we are entering a phase where AI begins not only to "think" but also to "act" in the physical world. The compactness and efficiency of the code created by Claude indicate that neural networks can optimize tasks at a level unattainable by humans. However, the problem of precise physical interactions remains the last bastion separating us from full-scale robotics automation. The next step is the integration of sensory feedback, and here we will likely see a new wave of competition between Anthropic and other market players.