The AI model Claude has surpassed humans in controlling a robot dog: speed is 20 times higher
Anthropic has presented the results of the second phase of the Project Fetch experiment, and the results are impressive. My team analyzed the data: the Claude Opus 4.7 model handled the setup and control of a four-legged robot 20 times faster than teams of human engineers. This is not just another benchmark—it is a demonstration of how AI is transitioning from the digital space into the physical world.
In August 2024, the experiment looked different: employees with no robotics experience used AI as an assistant to speed up finding solutions. Now, Claude Opus 4.7 worked almost autonomously, under minimal researcher supervision. The neural network independently completed a full cycle of tasks:
- connected to video sensors and LiDAR;
- wrote a program for manual control;
- created a robot trajectory monitoring system;
- configured an object recognition algorithm.
The performance figures deserve special attention. Opus 4.7 turned out to be 18 times faster than a team using previous AI versions, and 37 times faster than people working without chatbot assistance. Moreover, the neural network generated code that was 10 times more compact than human solutions. This indicates that the model not only speeds up the process but also optimizes it at a fundamental level.
Key takeaway: progress in robotics has become a side effect of the general scaling of language models. Anthropic did not introduce specialized algorithms for controlling hardware—the model adapted to the task on its own. However, there are limitations. Claude successfully guided the robot to the goal but could not neatly push a ball to the right spot. This requires complex real-time feedback, where humans still hold an advantage.
According to Anthropic's estimates, the industry is entering an era of "physical AI agents." In the future, neural networks will be able to use standard tools and equipment as effectively as they currently work with software code. In my view, this is a transition from narrowly specialized robots to universal AI-driven systems—and we are only witnessing the beginning of this journey.
My analysis: Claude Opus 4.7's ability to work with physical objects without specialized training is a signal that language models are becoming a platform for general artificial intelligence. However, issues with precise manipulations remind us that we still have a significant gap to bridge between digital and physical perception before achieving full autonomy in the real world.