Claude AI beats humans in controlling robot dogs: a breakthrough 20 times faster
The world of cryptocurrencies and high technology is accustomed to surprises, but news from Anthropic prompts reflection on fundamental shifts in automation. The Claude Opus 4.7 model, as part of the updated Project Fetch experiment, demonstrated capabilities surpassing human ones: it configured and programmed a four-legged robot dog 20 times faster than a team of experienced engineers.
Autonomy Without Limits
In August 2024, Anthropic employees with no robotics experience tried to make a robot move using AI only as an assistant. Now, Claude Opus 4.7 acted almost independently, under minimal researcher supervision. The neural network completed a set of tasks:
- Connected to video sensors and LiDAR;
- Wrote a program for manual control;
- Created a trajectory monitoring system;
- Configured an object recognition algorithm.
The results are impressive: Opus 4.7 was 18 times faster than a group using older AI versions, and 37 times faster than people working without chatbot assistance. A key factor is code efficiency: the volume written by the neural network turned out to be 10 times smaller than that of human teams, reducing the likelihood of errors and speeding up execution.
Side Effect of Scaling
The experiment's authors emphasize that this progress is not the result of targeted work on robotics, but a side effect of the general scaling of language models. Anthropic did not introduce specialized algorithms for controlling hardware, making the achievement even more significant. This confirms the thesis that universal AI systems can adapt to new tasks without additional training.
Limitations of the Physical World
Despite the success, Claude still faces difficulties with precise physical actions. The model managed to guide the robot to a target but failed at the task of gently pushing a ball to a specific point. This requires complex real-time feedback, where humans still maintain an advantage. An error in controlling a physical object is a reminder that the virtual mind and the real world are separated by a gap that remains to be bridged.
Anthropic is confident that 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 code. However, as the case of suspending access to Fable 5 and Mythos 5 models due to export controls shows, the development of such technologies goes hand in hand with geopolitical risks.
Expert Opinion: This experiment is not just a demonstration of speed, but a signal that AI is beginning to move beyond digital environments. For the crypto industry, where automation and smart contracts have already become the norm, the emergence of "physical agents" could open new horizons—from autonomous management of mining farms to logistics for DeFi projects. However, until AI learns to "feel" physics, manual intervention will remain necessary.