AI agent Claude Opus 4.7 surpasses humans in controlling a robot dog: a breakthrough in physical AI
The world of robotics has witnessed a landmark event: Anthropic's Claude Opus 4.7 model completed a complex cycle of configuring and programming a four-legged robot dog 20 times faster than a team of experienced human engineers. This refers to the second phase of the Project Fetch experiment, the results of which demonstrate a qualitative leap in AI autonomy.
In the first phase, which started back in August 2024, the neural network acted only as an assistant for people with no experience in robotics. Now, the situation has changed dramatically. Claude Opus 4.7 operated almost completely autonomously, with minimal supervision from a researcher.
Autonomous Assembly and Control
Without human intervention, the neural network performed a full range of tasks:
- Connected to the robot's video sensors and LiDAR;
- Wrote a program for manual control;
- Created a trajectory monitoring system;
- Configured an object recognition algorithm.
The model's performance proved impressive: it was 18 times faster than a team using older AI versions, and 37 times faster than people working without chatbot assistance. Moreover, the code generated by the AI turned out to be 10 times more compact than human-written code, indicating its high efficiency and optimization.
Physical AI: Progress Without Specialization
The key takeaway from the experiment is that Anthropic did not introduce specialized algorithms for controlling the hardware. According to the team, all progress in robotics is a side effect of the general scaling of language models. This confirms the thesis that a universal AI can adapt to new physical environments without retraining.
However, it is worth noting that, despite its success in navigation, Claude still struggles with precise physical manipulations. The model managed to guide the robot to its target but failed at the task of gently pushing a ball — a task requiring complex real-time feedback, where humans still maintain an advantage.
Anthropic believes the industry is entering an era of "physical AI agents," where neural networks will control not only code but also real tools and equipment.
Expert Opinion: The progress of Claude Opus 4.7 is a clear signal that the boundary between the digital and physical worlds is blurring. However, the "blind spot" of precise motor skills remains a serious challenge. Until AI learns to sense physical contact as well as it senses context in text, the full replacement of humans in manual labor will remain a matter for the future. The incident involving the suspension of access to Fable 5 and Mythos 5 models due to US export controls only underscores how strategically important this technology is becoming.