Claude AI Surpasses Humans in Controlling Robot Dogs: Breakthrough or Temporary Anomaly?
The artificial intelligence market continues to demonstrate exponential growth in capabilities. This time, Anthropic has unveiled updated results from the Project Fetch experiment, which challenge the traditional superiority of humans in engineering tasks. The Claude Opus 4.7 model completed a set of tasks for configuring and controlling a robot dog 20 times faster than teams of experienced human engineers.
It is worth recalling that back in August 2024, company employees with no experience in robotics attempted to program a four-legged robot. At that time, AI only acted as an assistant, accelerating the search for solutions. However, in the new testing phase, Claude Opus 4.7 worked almost autonomously under minimal researcher supervision.
Autonomous Operation of the Neural Network
During the experiment, the model independently performed a number of critically important operations:
- connecting to video sensors and LiDAR;
- writing a program for manual control;
- creating a system for monitoring the robot's route;
- configuring an object recognition algorithm.
The results are impressive: Opus 4.7 turned out to be 18 times faster than a team using previous AI versions, and 37 times faster than humans working without chatbot assistance. Moreover, the neural network generated more efficient code — its volume turned out to be 10 times smaller than that of human teams. This speaks not only to speed but also to the quality of solutions.
It is important to note that, according to the developers, progress in robotics has become a side effect of the general scaling of language models. Anthropic did not introduce specialized algorithms for controlling hardware — this happened naturally during the training process.
Not All Smooth: Physical Limitations
However, despite impressive successes in programming, Claude still experiences serious difficulties with precise physical actions. The model managed to guide the robot to its target but failed to carefully push a ball to the right spot. This requires complex real-time feedback, where humans still maintain superiority.
Anthropic believes 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. Let me remind you 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 control regulations.
My analysis: This experiment is not just a technical curiosity, but a clear signal that the boundaries between the digital and physical worlds are blurring. However, investors and developers should remember: while AI brilliantly handles logical and algorithmic tasks, real "physical interaction" remains its Achilles' heel. This creates both risks of overestimating capabilities and enormous opportunities for those who can combine the power of neural networks with precise mechanics.